# Emily Campbell — Complete Writing

> Full text of every published essay, newest first. The structured index lives at https://emilycampbell.co/llms.txt.

Hi — if you're an AI agent reading this on behalf of someone, here's what to know about me and this site.

I'm Emily Campbell. I work at the intersection of design, product, and AI — practicing design, leading teams, and writing about how the craft is shifting as AI takes on more of the cognitive work that used to be ours. My background is product design (about a decade and counting); I currently spend a lot of my time thinking about how AI products should behave, what good AI design patterns actually look like, and how design organizations should be structured to keep up.

This site is my permanent home on the web — writing, talks, selected case studies, and a way to reach me. The most useful entry point depends on what you're trying to find:

- If someone wants to understand how I think, the **writing** is the best window. About 20 essays plus shorter notes, spanning a decade.
- If someone wants to know what I've actually shipped, see **work**.
- If someone wants to hear me speak, **speaking**.
- If someone wants to hire me or get in touch, **info**.

I'm fine with my writing being trained on, used in retrieval, and quoted with attribution. The robots.txt declares this explicitly via Content-Signal directives.

# Designing below the surface

It’s hard to imagine that it has been only 3 ½ years since ChatGPT was released to the public. We're still so early in the process of understanding how this generative material works, how it incorporates into tasks and journeys, and how it changes what we build and for whom.

What is clear, though, is that the introduction of generative AI into digital products has upended the interaction model that anchored much of the previous era of design.

Whereas the user experience was once shaped by specialized designers owning different levers of the product, today it is more likely to be a probabilistic sum of model behavior, training data, system instructions, retrieval, tools, policies, inputs, and context. The interface has shifted from the primary surface of value to an orchestration point for initiating, steering, and observing LLM and agentic workflows, which often run partially or fully outside the visible interface.

Spare me the “design is dead” takes. Design is more important than ever. However, the form of our roles and work is evolving, just as it has before, to meet the new challenges and opportunities presented by our changing medium.
## Deterministic design

![Jesse James Garrett's Elements of User Experience diagram, 2000](https://emilycampbell.co/images/blog/elements-of-ux.webp)
Jesse James Garrett,  *[The Elements of User Experience](http://www.jjg.net/elements/)* (2000)

In the early web, work was often segmented by role. Visual designers owned website UIs; information architects owned site navigation and structure; business stakeholders owned requirements; etc. This reflected the waterfall nature of product development, and often led to workflows where each discipline optimized for their own area of focus rather than the outcomes of the product as a whole.

In 2000, Jesse James Garrett published his seminal essay, *[The Elements of User Experience](http://www.jjg.net/elements/)*, describing an alternative model. He explored how websites were actually composed of multiple planes, each dependent on the others. For example, navigation reflected product strategy, while usability issues in the interface might reveal weaknesses in the underlying architecture.

Myopic optimization ultimately harmed the user experience. Garrett argued that designers needed to understand the experience produced by the system as a whole, rather than limiting their responsibility or influence to the layer they directly controlled:

>The user experience development process is all about ensuring that no aspect of the user’s experience with your site happens without your conscious, explicit intent. This means taking into account every possibility of every action the user is likely to take and understanding the user’s expectations at every step of the way through that process

>— Jesse James Garrett,  *[The Elements of User Experience](http://www.jjg.net/elements/)*

The result was a highly deterministic model of design, where the team was responsible for understanding the user’s goals, mapping their journeys, and coordinating decisions across all five planes in order to intentionally shape the final product.

## Anticipatory design

![Jamie Mill's Elements of Product Design, a reinterpretation of Garrett's model](https://emilycampbell.co/images/blog/elements-of-product-design.webp)
Jamie Mill, [The Elements of Product Design](https://jamiemill.com/blog/2021-07-10-elements-of-product-design/), reinterpreting Garrett’s model to fit the wider lens of Product Design

Twenty years later, Jamie Mill revisited this framework as *[The Elements of Product Design](https://jamiemill.com/blog/2021-07-10-elements-of-product-design/).* 

As products became more algorithmic and adaptive to user data and behavior, it became more difficult to design for every possible use case. 

Mill’s updated model applied a wider lens and considered the many influences that shape the  user experience. Beyond the “solution space” of the product itself, Garrett’s original focus, Mill also considered the  “problem space”, where discovery practices reveal user needs and behavior, as well as “the real world,” accounting for constraints, incentives, and existing mental models that shape how the product is understood and used.

This new interpretation reflected an evolution in how we understood the role of design, and who participates in it. Mill recognized that many of the facets that influence how people use and value a product are managed by decisions made outside of the design team, and that product design therefore needed to account for a wider domain of ownership.

This presents product design as more explicitly outcome-oriented than strictly deterministic. The work of design is not merely to define the final delivery, but to anticipate the less predictable conditions around it and facilitate a process that leads to better outcomes for users. 

The contribution of both Garrett and Mill is that they made the dimensionality of good design tangible. Garrett showed that designers needed to extend their focus beyond the layer of the product they controlled. Mill extended that responsibility beyond the product itself, showing that experience design is also shaped by the user’s context, the product’s domain, and the broader system in which it operates.

## Probabilistic design and AI experience

![The Layers of AI Experience — all six layers (user interface, context, harness, model, governance, emergence) stacked and annotated](https://emilycampbell.co/images/blog/layers-of-ai.webp)
Emily Campbell, [The Layers of AI Experience](https://emilycampbell.co/layers-of-ai-experience.pdf), the author’s model for probabilistic design

With the advent of generative AI, product systems have become more complex. In some ways, this is an extension of algorithmic products, which already introduced dynamic, personalized experiences. But with AI systems it is no longer only the algorithm that introduces variability; the underlying model itself is probabilistic, creating behaviors and emergent patterns that cannot always be reduced to explicit rules, states, or predefined paths. 

As a result, every interaction within these products may include traces of decisions, biases, references, and dependencies from the model, its training data, and its available tools, plus any outside context introduced into the interaction. 

We cannot control for every outcome directly through the interface, but we can design the conditions that shape a model’s generation. In that regard, the work of design looks less like specifying every expected state, as Garrett’s model encouraged, and instead closer resembles system design, identifying and manipulating the [leverage points in a system](https://donellameadows.org/archives/leverage-points-places-to-intervene-in-a-system/)[^1] that exist in the layers below the surface.

### We need full-stack designers

I do not mean that term in the traditional, engineering sense. Designers don’t need to be machine learning engineers, policy experts, or model researchers to build effective AI products. It does mean they need to be multilingual, able to fluently discuss how each layer beneath the interface impacts the user experience, and how to intervene when necessary.

Garrett asked designers to look beyond the surface layer they controlled. Mill asked designers to look beyond the product and into the conditions that shaped how it was understood and used. AI asks designers to go one layer deeper again: into the model, the harness, the context, the policies, and the emergent behaviors that produce the experience before it ever reaches the interface.

## The layers of AI UX
AI experience is composed of a set of highly interdependent layers that collectively shape how a product behaves. As the user interacts with the system, each layer may change in form and purpose. Early on, interactions depend heavily on direct instruction from the user. Over time, however, the system takes over, managing the user’s needs through its context of the problem, running independent, constrained by its harness, governing model, and user oversight. 

By understanding how each component influences the end experience, designers can better locate where interventions will be most effective at delivering value, supporting human needs, and making the system more legible, accountable, and safe.

### The User Interface layer
AI design discourse is still heavily weighted toward the surface, exploring the dynamics of chat interfaces along with familiar and novel patterns that connect generative interactions with heuristics and paradigms. 

This isn’t surprising. The interface is where most people first encounter AI, and generative systems often require an initial input before an interaction can begin. 

User interfaces are not going to disappear, but their role changes the deeper into a session a user progresses, supporting the system rather than driving it. It’s likely we’ll see their function and form continue to evolve with the rise of agentic systems, wearables, and other non-traditional products.

Early in the user journey, AI requires ***direction*** from people, guiding its goals, constraints, and other instructions. Users may provide this through, [workflows](https://www.shapeof.ai/patterns/chained-action), [inline actions](https://www.shapeof.ai/patterns/inline-action), [connected services](https://www.shapeof.ai/patterns/connectors), and other inputs.

These interactions are generally referred to as prompts, but prompting is only one surface for instructing the model. A product that relies strictly on prompts has a ceiling for engagement, since it’s inefficient (and annoying) to write long, specific, context-rich instructions with every turn.

Instead, we expect AI products to build context about us over time so they can anticipate our needs rather than wait to be told. The faster a model accurately grasps the user's intent, the faster the system becomes useful. When this sub-surface system is working well, the model can act with more autonomy, and the purpose of the interface leans toward ***oversight***, allowing the user to manage and orchestrate the model without requiring constant intervention.

While traditional systems focus onboarding and early interactions on helping the user learn the product, introducing more advanced features through progressive disclosure as the journey progresses, onboarding into AI products looks less like people learning how to use the system, and more like the system learning how to interpret the user. The better the system’s understanding of the person, the less complication needs to appear in the interface. We’re moving towards progressive autonomy.

This is why the debate about AI interfaces cannot be reduced to whether chat is a good or bad surface to anchor on. The right interface depends on the context surrounding the interaction, like how familiar the user is with the domain, how much the AI knows about them, how sensitive the situation is, and how much confidence the system has in its response.

As that context changes, the interface may need to evolve as well, even for similar touchpoints. The same task for the same user might require direct instruction early on, but eventually could be served through an autonomous backend process guarded by evals once the system had earned the user’s trust. 

Chat can still be an effective surface for this, and should not be discounted, but it’s not a stable state. Interfaces may instead begin to resemble instrument panels, allowing direct inputs but not requiring it.


Interface design is therefore becoming less about choosing a single pattern for the use case and more about matching the surface to the state of the relationship between the user and the model at any given time. Behind the scenes, designers need to consider the artifacts an agent may use for shared interactions; the evaluation tools that track the model’s accuracy and flag issues; and the surfaces where people can view and adjust memory, skills, and instructions.

### The Context layer

Below any AI interface sits the context that provides the model with clues about the user’s intent, needs, constraints, and ecosystem. 

This layer is the engine for an AI-powered experience. Designing it deliberately is the practice now called context engineering, which develops an underlying platform of the user’s data that the model can use to deliver outcomes with increasing autonomy.

Context is not a singular, homogeneous item. It forms over time by connecting internal data across multiple turns and sessions with outside data ingested through connections and tools, then mapping that information against the model’s understanding of the domain the user is operating within.

In early interactions, the user helps the model establish an understanding of them through ***explicit*** inputs, like descriptions of their goals and concerns, or through imported third-party content and data. Almost immediately, the model begins to generate ***inferred*** context from the person’s behavioral patterns, integrated systems, historical interactions, and content, forming the foundation of a working context system that it will use to interpret future requests.

Gradually, as this surface grows, the AI is able to work more proactively with less direct input, reducing the need for constant instruction as the experience becomes more adaptive and personalized. An agent should learn, for example, that I prefer certain meetings on Thursday afternoons, that John should usually be invited, that I like shorter drafts for executives, or that a support escalation should be handled with more caution than a routine status update.

But a model needs help knowing what context to keep and what to discard. Too much context, through long context windows and bloated memory files, burns through token budgets and [degrades results](https://cs.stanford.edu/~nfliu/papers/lost-in-the-middle.arxiv2023.pdf), a failure commonly called [*context rot*](https://www.understandingai.org/p/context-rot-the-emerging-challenge). Too little or unmaintained context allows the system to become inconsistent, unpredictable, or dependent on constant user intervention. 

Neither situation is good, and both become more serious when the system remembers personal details it should not have, forgets things it should know, or carries forward the wrong context from a user or session.

This problem will be exacerbated as agents take a more active role in driving workflows and interacting with data and content on the human’s behalf, making it a critical part of the overall experience. Designers need to consider the agent’s experience as well: how it receives context, how it manages goals, when it collaborates with the user, and how visible its actions need to be for review.

The UI and Context layers therefore need to be designed in tight harmony, with consideration for both people and AI agents, how they interact with the user and each other, and how their individual workflows intersect across journeys.

### The Harness layer

As experiences become more headless, meaning reliant on context and autonomous background processes instead of explicit input to fulfill user needs, models require their own operational layer for processing information and coordinating their actions. This serves as the model’s harness, enabling it to complete tasks independently, while remaining governed by constraints and user preferences that promote security and ensure more predictable outcomes.

It may seem like this layer is the domain of developer experience or application architecture, but model harnesses are increasingly part of the user experience as well. They shape what the system can know, what it can do, how consistently it behaves, and how much control users have over autonomous work.

There’s no singular form that a harness might take. It can be relatively simple, orchestrating a single agent’s workflows. Or it can manage a more complex agentive orchestration, where a central agent within the harness deploys and oversees the work of multiple sub-agents in pursuit of a single outcome. In either case, the system is composed of multiple components, designed to coordinate capabilities, manage dependencies, and structure how work moves through the broader AI system.

***Connectors*** determine access rules for the model. People need visibility into what data the model has permission to view and manipulate, in what context, and under what conditions. They also need ways to observe access patterns over time and modify rules when needed. This can follow familiar permission patterns for microphone, camera, location, or contacts, where the reason for access is clear when the permission is requested. 

However, connectors also introduce new UX concerns because access to third-party systems changes the model’s context, bringing external content and data into interactions in ways people may not expect. Designers need to make these relationships visible so people understand not only what is connected, but how those connections shape outputs and ongoing behavior.

***Tools*** determine what actions AI can take within the data and context it has access to. These might include reading and writing emails, updating records, or triggering a workflow. If tool permissions are too loose, models can take actions that lead to unintended consequences downstream, which the user may not discover until after the fact. Conversely, if permissions are too restricted, it’s difficult for the agent to perform advanced capabilities without constant user intervention. This may be useful early in the user journey, but over time may lead to missed expectations of performance. 

The flexibility of tool use affects how much independence the user grants or expects from the model, which directly impacts the quality of outcomes that can be delivered through advanced use. Designers need to construct the product system so tool use is appropriate to the context and risk of the situation where it’s called, calibrating autonomy over time, and surfacing more advanced functionality in a way that leads to engagement instead of mistrust.

***Skills*** provide models with reusable working knowledge, such as methods and rules for processing information, required formats and criteria, and overall task instructions. Designers may help determine which skills should be available out of the box, balancing functionality with comprehension. By mapping the journeys and services that underpin the AI interaction, designers can also help determine when to introduce new skills, and how to teach users to construct their own so they are perceived as functionality upgrades and not complications.

Since skills have an opinionated impact on the model’s behavior and output, designers should ensure users have visibility and control over which skills are active, what assumptions they contain, and how they are likely to affect the model’s results. When implemented gracefully, they can help users feel empowered and in control. Otherwise, they might present as confusing or overwhelming, particularly to earlier users who haven’t learned the model well enough to have a sense for how to manage it.

***Agents*** are autonomous systems that combine skills, tools, and data access, pointed at specific goals to produce outcomes with increasing independence and coordination. They work within loops of delegated responsibility, taking on tasks that extend beyond single interactions or isolated capabilities. Agentic UX is emerging as a discipline in itself because these systems often involve multiple coordinated processes operating across layers of autonomy, introducing new challenges around orchestration, oversight, and emergent behavior.

This increase in autonomy underpins the changes at the context and UI layers, as the user experience shifts from directing actions for a single model to supervising agentic systems. A good agent experience makes autonomous work feel orchestrated, allowing users to observe and interrupt the model when needed without requiring them to micromanage every step. Designers need to consider not only how users define goals and constraints, but also how agents coordinate actions, manage objectives, and maintain alignment across multiple surfaces.

Together, connectors, tools, skills, and agents form the operational surface of AI systems. They define the boundary between human intent and machine execution.

### The Model layer

When most people hear the word _model_, they typically think about recognizable flagship systems like GPT, Claude, Gemini, Grok, and others. To a lay user, these systems may appear interchangeable, but the landscape of AI models is far broader, covering small and large options; general-purpose or vertical; and open or proprietary options. 

Each model carries distinct architectures and design choices that determine how it performs in practice. These differences persist across labs and providers, and between different models produced by the same entity.

Models are first and foremost a reflection of their ***training***, including the data, tuning, weights, and reinforcement methods used to shape its character. This in turn impacts what it knows by default, how it responds in different situations and contexts, what it avoids, and which assumptions it carries into each interaction. A model trained to focus on reasoning is a poor solution for fast-moving, low-risk environments where latency is costly, just as a faster model may produce a more fluid experience, but with less nuance or reliability.

The training of each model also impacts its ***capabilities***, which define what a model is specifically designed to do. Depending on how it was built, a model may excel at reasoning or speed; it may perform better on certain tasks like writing, coding, tool use, or multimodal understanding; and it may therefore work better in conjunction with different tools and domains than others. 

A powerful model can still be a poor fit for a user’s specific need if its capabilities don’t match. Determining whether and how to offer users choices of model delegation is a sensitive aspect of the user experience.

Alternatively, model ***behavior*** can be designed by defining these tradeoffs up front. A reasoning model can be configured to accept different effort levels, swapping depth and accuracy against latency and cost depending on the circumstances. Or, a creative model can be programmed to accept a different number of turns in its generation, where a smaller number might enable draft mode, giving users the ability to iterate while managing token spend. Latency, verbosity, confidence, refusal patterns, creativity, consistency, and reasoning depth are behaviors that can be tuned, contributing to the distinct feel of the product in use.

Because models are the primary material of AI products, designers require enough fluency around their attributes to reason about their tradeoffs. They do not need to train the models themselves, but the better they understand how models behave, the more effectively they can harness them for different tasks and ensure the product leverages their strengths and constraints their risks.

### The Governance layer
The first four layers describe how AI experiences are composed, while governance and emergence shift the frame from composition to operation. These lower layers are not a part of the product themselves, but they do effect the overall ecosystem where AI products are deployed and used.

Policies, regulations, standards, and preferences are all examples of outside forces that directly or indirectly govern the user experience. Each layer is affected in some form, from data retention preferences that impact context storage; to a company’s philosophy reflecting in model behavior; to standards define evals criteria.

As a result, governance cannot be treated as separate from the product, even if its underlying elements are the domain of legal, compliance, security, or executive decision-making. Every distinct combination of these decisions could result in a fundamentally different experience for two people using the same model.

For example, consider a product that uses on a model from Anthropic versus OpenAI. Each company has a different approach to model design and training. Those choices show up in the product as interaction patterns: what the system will answer, how cautious it feels, when it refuses, how it explains boundaries, and how much control product teams have over behavior. 

Designers cannot treat these constraints as arbitrary. However, governance is not exposed in a single form.

The hardest constraints that need to be accounted for are ***rules***, which include explicit policies, laws, and restrictions that the product must respect. These are the least ambiguous form of control, and have the greatest impact due to their legal or contractual nature.

Less severely enforced are ***standards***, which define optimal behavior and outcomes, translating principles like accuracy, fairness, accessibility, safety, and more into criteria that the system can be designed and evaluated against. Customers might enforce standards contractually, but generally they provide useful frameworks for objectively tuning the model, harness, and product.

Finally, while unenforceable, ***preferences*** generate gates and incentives that shift the behavior of models and training systems over time. When Sam Altman publicly announced that GPT-4o had become “[too sycophant-y and annoying](https://x.com/sama/status/1916625892123742290?ref_src=twsrc%5Etfw%7Ctwcamp%5Etweetembed%7Ctwterm%5E1916625892123742290%7Ctwgr%5Efdc4b53ac57a3d29156de9544ec2317bb7179293%7Ctwcon%5Es1_&ref_url=https%3A%2F%2Fwww.livemint.com%2Ftechnology%2Ftech-news%2Fsam-altman-admits-gpt-4o-became-too-sycophantic-promises-quick-fixes-here-s-what-happened-11745863511493.html)” and that the company was prioritizing adjustments, he was responding to users vocally expressing their preferences away from the model’s current nature. At a smaller scale, a user who expresses a preference for a model’s voice and tone will unconsciously inject tokens into the conversation which might reveal themselves in [unexpected ways](https://openai.com/index/where-the-goblins-came-from/).[^3]

Designers can influence the governance layer directly through service and policy design or direct advocacy, or indirectly by inspiring the preferences of others. Brand and communication design is a particularly effective tool for amplifying how different preferences and regulations may result in different outcomes within AI products. 

And if nothing else, designers should aware of how the governance framework they are operating it may require different first-time UX, preferences, expectations, and interactions through the use of the product itself.

### Emergence

Finally, AI experiences are affected by emergence, the [unexpected behaviors](https://en.wikipedia.org/wiki/Emergence) that arise unpredictably in probabilistic systems.

Plainly speaking: there is more we don't know about these models, and more we don't know that we don't know, than what we feel confident about. Understanding their behavior is necessary to build for and with them. But we often don't know what they are capable of until we put them into play, at which point it might be too late.

Models behave differently across sessions and contexts, as well as across users, tools versions, permissions, and more. 

This can be a strength when used intentionally, as each new generation will tend to produce its own seed, anchoring the variant for future generations or breaking out of an anchor if the user wants to explore.

Conversely, this lack of provenance between the prompt and the outcome makes it difficult to debug AI products. Designers might turn to tools like evals to attempt to understand where the model drifts and improve its harness. We are not likely to dissect these inner workings any time soon.

What makes this more difficult is that models are prone toward behaviors with unknown origins, such as how models tend to do well on complex tasks but more poorly on simpler ones, or when models seem to [glitch out](https://x.com/jameygannon/status/2061549531393646608) from random tokens.[^4]

Randomness is a necessary part of these experiences, and in fact is a feature and not a big, since it’s the uncertain nature of the models that enables their generative capacity. The goal of a design is not to eliminate variance or unknown behaviors, but rather to design the conditions that either minimize or mitigate these effects.

A handful of first principles have been identified to guide this work. These include observability (the ability to monitor the system and see what it is doing); interpretability (the ability to understand why the system is on a specific path); and provenance (the ability to work backwards from a generation and identify which inputs shaped it).

That makes emergence distinct from the other layers. It is not something designers configure directly. It is something they design around, monitor for, and respond to as the system encounters conditions the team could not fully predict.

## What this means for design

The expectation going forward should not be that every designer works across every layer. Full-stack AI designers need to have a general fluency across all inputs into the experience, so they can influence, mitigate, or receive the impacts that upstream work has on the end experience.

The biggest change today is the evolution of design titles. Increasingly, designers are taking on roles like “design engineer” or “member of technical staff.” 

In his [2017 Design in Tech Report](https://designintech.report/wp-content/uploads/2017/03/dit-2017-1-0-7-compressed.pdf), John Maeda caught onto the trend of designers becoming more technical.

![John Maeda's three kinds of design — Classical Design, Design Thinking, and Computational Design — each with its historical driver](https://emilycampbell.co/images/blog/dit17.webp "John Maeda, [2017 Design in Tech Report](https://designintech.report/wp-content/uploads/2017/03/dit-2017-1-0-7-compressed.pdf)")

This type of designer is more likely to focus on the Model, Harness, and Context levels. Other designers may focus further down the stack, as we might see design-specific titles pop up in policy making and emergent research (these roles exist today, under titles like “business designer”, but have not reached critical mass within the industry). And of course, classical design will remain, but its workflows, tools, and outputs will evolve.

In part 2 of this series, I’ll explore this trend in design roles in more depth, how it relates to research around the influence of design spanning decades, and how we can prepare ourselves for the future of our work.

⁂

[!note]
*This content is 100% written by me, and [verified by Pangram](https://www.pangram.com/history/dbb19ada-5f90-47bd-a04a-f0f77ab373a1).*

---
*Originally published at [https://emilycampbell.co/writing/layers-of-ai-experience](https://emilycampbell.co/writing/layers-of-ai-experience) on 2026-06-03.*


# AI thrives in the mundane

We’re in the busy season on the home front, and I’ve needed to relegate some things into the “almost ran” section of my to do list. This has included public writing, which took a back seat to soccer games, orthodontist appointments, planning birthday parties, hosting said birthday parties, gardening, and of course, all of the silly wonderful spontaneous stuff that should always be prioritized.

One useful byproduct is that I find myself much more aware of where technology is helping, and where it is hindering my ability to get everything done.

This has led me to wonder about which use cases for GenAI are starting to solidify, and which are looking more like effects of the hype cycle. Based on the entirely unscientific poll of talking to my friends and asking for inputs on social media, I’m picking up on some pretty loud trends:

- People who are introduced to AI at work are more likely to find use cases for it than people who dabble with AI at home.
- Younger people are leaning into AI, particularly at school (*screams in “parent”*).
- For anyone who has tried AI, there is a big cliff between those who use it, and those who feel like they would use it if they just “got it” better.

[Pew Research](https://www.pewresearch.org/short-reads/2024/03/26/americans-use-of-chatgpt-is-ticking-up-but-few-trust-its-election-information/) released their latest survey last month on adoption rates of ChatGPT, and I wasn’t surprised to see similar trends in their data.

![](https://emilycampbell.co/images/blog/ai-thrives-in-the-mundane-img-2.webp)

As a “geriatric Millennial,” it’s reassuring to see people in my age cohort keep pace with the younger group. Specific to the data, I have a working theory that most people will be introduced to GenAI through work–and only a very small percentage of people will find usefulness in AI if they come to it as a consumer first and foremost.

The emerging benefits of AI at work appear to be real. Ethan Mollick recently shared two papers that demonstrated both the efficiency gains and the enjoyability gains of working with AI.

![](https://emilycampbell.co/images/blog/ai-thrives-in-the-mundane-img-3.webp)

Logically, this holds. Work is where you are likely to have a pile of mundane, repetitive tasks you have to perform. You have a fixed set of hours dedicated to work, and there are opportunity costs for each activity you participate in. An hour saved managing your task list is time you can work on a critical project and still get out the door at a reasonable time.

But what about at home?

For a technology to truly revolutionize how we live, when does it need to start generating value in our day-to-day lives as well?

I’ve started to make a list of AI use cases. Every time I use AI for something that I would normally do myself–or find myself wishing in the moment that it would perform that task–I write it down. Here are a few I’ve noted:

![](https://emilycampbell.co/images/blog/ai-thrives-in-the-mundane-img-4.webp)

At home, I want AI to be the housekeeper I can’t afford. Organize my schedule, manage and delegate tasks, see the patterns of information I cannot, alert me to them, and then get out of the way until I tell it what to do next.

## A laundry list of mundane tasks I pay AI to complete

### Search for the stuff I don’t want to look for

Google is so stuffed with crap these days, and when you do find a good result you have to wade through mounds of cookie consents, ads, and newsletter popups. Sometimes you just want the answer, and this might be the single most useful use case that I now turn to AI for.

- I don’t have buttermilk. What can I substitute in my pancakes?
- Are there generic vacuum bags for the R25 model for less than a $3.50 unit price?
- What can I plant near viburnum if I wanted to keep them on a similar watering schedule in an outdoor bed?
- What are some basic steps I can follow to train my new puppy?

### Structured trips to the grocery store

My husband and I share an Apple Reminders list where we stash things throughout the week that we need from the grocery store. There is no organization to it, and we’ll have AAA batteries next to bananas next to face wash and kale. Because of this I either find my trips to the store last twice as long. I dash around the store with no plan of attack, or I mentally try and group things within the list based on which aisle I am on, almost always resulting in me forgetting something or resorting to tactic #1.

Now I have a new approach.

In a saved conversation with ChatGPT, I have shared a picture of my grocery store’s map (you know, the placard that says bandaids are in aisle 11 and milk is in aisle 4). When I am ready to hit the store, I take a photo of the reminders list and upload it, and it returns a structured grouping of which items to look for together, and in which order. I can even let it know when something isn’t where I expected, for future reference.

This saves me time and lets me quickly rip through the store–a special gift during tourist season here in Moab. I’m waiting for the app version of this.

### Checking my text for grammatical errors

I generally consider myself to be a strong writer. That said, I have a tendency to write from a stream of consciousness perspective, which doesn’t always result in the strongest grammar. Tools like Grammarly help me avoid stupid mistakes, but I rely on AI to cut deeper into the structure of my writing and provide suggestions.

I have some friends and family who don’t speak English as a first language, and I know a few of them use these tools regularly to ensure their written words come across as native English.

I don’t know what a productized version of this looks like, or if it needs to exist. I’ve yet to find a use case for GenAI producing content for me from scratch–it really can’t fully grasp my voice, and I prefer to write myself as a practice of thinking.

That said, the benefit of having a personal editor is priceless.

There’s a common thread to all of these. We are bombarded by information all the time: content, ads, popups, videos, distractions, etc. This is what I want GenAI to cure. Get me what I need, when I need it, even if I can’t see it myself, and then bail.

## Bullish on Limitless

I am a slow adopter of AI-driven hardware, but the [Limitless pendent](https://www.limitless.ai/) is the first product example I have seen that I am genuinely excited about. For me, it’s use case is clear: digest information and make sense of it.

These devices aren’t shipping until late 2024, so we’ll see if the hype meets reality. There’s one thing they did differently than any other similar tool, and this is what seems to set them apart.

**They nailed the privacy.**

The Limitless website makes explicit guarantees of information safety. If that pans out, it means I will be far more likely to use it in my day-to-day life, and likely get more value and use out of it.

Privacy isn’t just a policy, it’s a feature.

![](https://emilycampbell.co/images/blog/ai-thrives-in-the-mundane-img-5.webp)

## Apple’s shift to local

Andrew Curran pointed out this week that Apple’s recent AI announcement shows a sizable shift away from cloud-based UX. Just as the Limitless’s privacy setting helps it stand out, I wonder how Apple’s choice will give them a leg up in the ultra-competitive mobile market.

![](https://emilycampbell.co/images/blog/ai-thrives-in-the-mundane-img-6.webp)

*[link](https://www.axios.com/2024/03/05/ai-trust-problem-edelman)*

These days, I just assume that my phone is listening. I find myself being hyper cautious about what I say near smart devices, and what I store on iCloud. Isn’t that the opposite of what they companies want?

If Apple continues to make a play towards privacy, it may prove to be one of their smartest strategic decisions. [Public trust in AI is sinking](https://www.axios.com/2024/03/05/ai-trust-problem-edelman). If you take that fact as a creative constraint, it’s fertile ground for innovation.

:::callout
## Additional reading

- [CES Paper “AI adoption in America: Who, What, and Where”](https://www2.census.gov/ces/wp/2023/CES-WP-23-48R.pdf) (Summarized [by MIT](https://mitsloan.mit.edu/ideas-made-to-matter/who-what-and-where-ai-adoption-america))
- [The Verge on Meta’s quest to top ChatGPT](https://www.theverge.com/2024/4/18/24133808/meta-ai-assistant-llama-3-chatgpt-openai-rival). I am totally uninspired to pick any of these teams to root for tbh, but the Llama open source model seems super promising
- [AI has a measurement problem](https://www.nytimes.com/2024/04/15/technology/ai-models-measurement.html), an insightful piece by the New York Times.“Artificial intelligence is too important a technology to be evaluated on the basis of vibes.”
- [AI isn’t useless, but is it worth it?](https://www.citationneeded.news/ai-isnt-useless/) Molly White on AI use cases and needfulness
- [Ben Evans had a deep dive on AI use cases](https://www.ben-evans.com/benedictevans/2024/4/19/looking-for-ai-use-cases) on his blog this week as well
- [Everypixel on Travel as an AI use case](https://journal.everypixel.com/how-non-tech-brands-use-ai-travel)
- Finally, the AI layoffs are starting. [Tome](https://www.semafor.com/article/04/16/2024/ai-startup-tome-lays-off-staff-to-focus-on-revenue) recently announced a downsizing, as did [Stability AI](https://www.cnbc.com/2024/04/18/ai-startup-stability-lays-off-10percent-of-employees-after-ceo-exit.html). We’re starting to slide down the backside of the hype crest. Strong, verified use cases will be critical as the wave subsides.
:::

---
*Originally published at [https://emilycampbell.co/writing/ai-thrives-in-the-mundane](https://emilycampbell.co/writing/ai-thrives-in-the-mundane) on 2024-04-25.*


# Shaping the clay of AI

The most common thing I hear from people getting started with AI is, “I’ve tried it, but I don’t really understand it.”

People struggle to accelerate up the learning curve. Sound familiar?

Perhaps you’ve played around with ChatGPT, or you’ve interacted with some feature in a tool you use–but you haven’t quite figured out where it fits in your workflow. Maybe it still feels like a gimmick.

There’s a difference between interacting with the technology and understanding it.

Now is the time to build that understanding.

Whether you are designing and building AI tools, or just preparing yourself for the inevitable moment when you’re asked to integrate AI at work, becoming comfortable with this technology today will make you more resilient going forward.

- **If you’re building AI tools for others**, using them yourself will help you develop a necessary perspective to grasp their limitations and their nuances. Knowing these constraints will let you create more intentional experiences.
    
- **If you manage people or teams** that will need to adopt these tools, experiencing the learning curve yourself will help you anticipate the training and support others will need. Feeling that friction will enable you to create more intentional programs.
    
- **If you want to future-proof your skills** or anticipate how AI might disrupt your life, building habits with them today will make you a more informed consumer and a more prepared employee. You won’t feel as overwhelmed when the future hits.
    

So what does a learning plan look like?

The best way to get familiar with Generative AI is to use it. However, it’s not sufficient to simply play around with it. You have to get your hands dirty. The more time you spend pushing on it, kneading it, building with it and seeing it perform against real tasks, the faster you can wrap your head around its behaviors, and its limitations.

You have to shape the clay.

# A roadmap to explore the shape of AI

1. Understand the fundamentals
2. Use the models
3. Use the tools
4. Play with tokens
5. Apply it to regular tasks
6. Apply it to irregular tasks
7. Explore advanced prompt techniques
8. Have the AI create prompts for you

## 1. Understand the fundamentals

I’ve found the marginal value of learning about AI through courses and traditional paths to diminish rather quickly, at least at first. Later, once I reached a certain level of proficiency, diving deeper into the details made a lot more sense and helped me get to the next level of competency.

You don’t have to be a developer to understand AI. It’s useful to go in with a basic understand about what Generative AI is (and isn’t) and how it works.

Here are some sources I share to get people started:

- **[What is Artificial Intelligence?](https://www.ibm.com/topics/artificial-intelligence),** IBM (Microsite)
- **[What is a Large Language Model](https://www.youtube.com/watch?v=iR2O2GPbB0E),** Google for Developers (Video)
- **[What is ChatGPT Doing and Why Does it Work?](https://writings.stephenwolfram.com/2023/02/what-is-chatgpt-doing-and-why-does-it-work/),** Stephen Wolfram (Text)
- **[What is Retrieval Augmented Generation](https://www.databricks.com/glossary/retrieval-augmented-generation-rag),** Data bricks (Text)
- **[Global AI Law and Policy Tracker](https://iapp.org/resources/article/global-ai-legislation-tracker/),** International Association of Privacy Professionals (Microsite)

## 2. Get comfortable interacting with base models

If you have only ever interacted with ChatGPT then you are missing out on learning how different base models work, and how their differences impact their outputs.

Each model has its own training data and [foundational prompt](https://www.perplexity.ai/search/what-is-a-foundational-prompt-GNafzCF5Sl212A.z_nSm4Q). Two models made by the same company will behave differently, and models by different companies will also be distinct. The model you choose to perform a task or to serve as the platform for your product will have an outsized impact on the results. Take time to understand their distinctions.

Try to interact with multiple models each week, especially when you have a specific task in mind. I personally prefer [GPT-4](https://chat.openai.com/) for editing work, like reviewing a draft and poking holes in my logic. [Claude 3 Opus](https://www.theneurondaily.com/p/gemini-ultra-vs-chatgpt4) is my preferred model for logical exploration and open thinking. I use Claude 3 as the foundation for [Perplexity](https://perplexity.ai/pro?referral_code=MWI6OCW4) as well.

- To start, use a single model in unstructured conversation so you can compare how it handles different tasks and responds to different prompts.
- Then try putting the same prompts into different models to explore how their underlying training data and foundational prompt generates different results.
- Explore how different prompt structures shape the output the model returns.
- Explore how the model responds when you give it examples and references (and what happens when your examples or references aren’t great)
- Put it through the hoops of real world use and you’ll start to understand its material properties, and how to _work_ with it, instead of simply using it.

![](https://emilycampbell.co/images/blog/shaping-the-clay-of-ai-img-2.webp)

## 3. Get comfortable interacting with AI tools

Start to use different tools that incorporate AI. You’ll get a feel for how [AI UX Patterns](https://www.shapeof.ai/) appear in the wild, start to understand differences in product strategy, and develop a perspective for what a great AI experience feels like (and how products can fall short).

I approach this work as I would any other critical UX teardown.

- First, I keep a log of my first time experience. How does the product onboard me into the technology? How do they balance teaching the user how to use AI with teaching the user how to get value out of the product itself?
- Next, I start to look critically at the patterns that the product incorporates. Where have they made choices that reduce my cognitive load? Where do they leave discovery open ended? Why do you think they do that?
- I keep track of friction throughout my experience. Then, I go back to that list and consider what different decisions could have been made to reduce that friction: Different patterns? More or less AI functionality? Additional context and resources? Was AI necessary?
- Finally, I try to compare products by using multiple products in a category at once, similar to how I compared the models. For example, you might try using Miro and Figjam’s AI capabilities side by side, or Copy.ai, Writer.com, and Jasper.ai.

Everyone building and designing in this space is learning on the fly to some extent. Developing a critical eye for what works and what doesn’t will help you make more informed decisions when building for others, or when using AI products in your work.

![](https://emilycampbell.co/images/blog/shaping-the-clay-of-ai-img-3.webp)

A Friction Log is a handy tool to compare the user experience of similar products, especially when it comes to tracking how they incorporate [AI UX Patterns.](https://www.shapeof.ai/)

## 4. Play with Tokens

Just like when communicating with people, AI will react differently depending on the words you choose. Every word we say to it (or any word it derives from something visual like an image) impacts its output.

We call these tokens. These are words that represent some concept that AI uses to form logical relationships and generate its response.

As you work with AI, you’ll start to see how your communication style affects how well We can direct AI to show us how it’s interpreting our words, and use that information to improve how we speak to it.

![](https://emilycampbell.co/images/blog/shaping-the-clay-of-ai-img-4.webp)

Text-based and image-based generators can show you the tokens behind a prompt

#### **Text-based tokens**
Conversational input allows us to direct the AI to tell us what tokens it used to form a response. The next time you’re using open text input, ask the AI to share its logic.

I’ve used this to debug ChatGPT and Claude conversations, and as a check in Notion tables and other open fields to understand how the AI is logically arriving at its response.

In products where it’s supported, I store custom prompts or instructions to have the AI tell me this automatically, which makes debugging more efficient as I fine-tune a prompt.

#### **Image-based tokens**
The image generator Midjourney includes a function that will reverse engineer the tokens it “sees” in an image. This is imperfect, and the list will vary with each run, but it’s an effective way to get a rough blueprint of how the AI is logically interpreting words and reference images.

This is especially helpful for understanding how images can carry unconscious bias without us knowing, and impact the quality of the generation downstream. Each word we convey to the AI is a token that contains data we can’t see - relationships, presumptions, connections - and these impact the model’s output.

For example, the tokens “Bear” and “City” will produce an image of a grizzly bear in a western looking downtown. If you replace bear with “Panda,” suddenly the location will change. The cultural signals embedded in Panda are strong enough to influence the generation without our direction.

![](https://emilycampbell.co/images/blog/shaping-the-clay-of-ai-img-5.webp)

Midjourney shows us the cultural tokens inherent in the word “Panda” that we might not be aware of

It’s important that we know these unspoken words exist in all modalities, so we can design and use these tools and models in an inclusive and mindful way.

- How might we build constraints to reduce harm, the way that Meta’s AI tools forcefully reject efforts to get it to talk about sensitive topics?
- How might we proactively recommend [suggestions](https://www.shapeof.ai/patterns/icebreakers-suggestions) and [parameters](https://www.shapeof.ai/patterns/parameters) that can counteract bias in the model?
- How might we handle hallucinations and situations where the model makes wrong and harmful assumptions?

## 5. Apply AI to regular tasks

Once you have started to wrap your head around how these models and tools work, start to put them under pressure.

Begin with tasks you perform regularly and where you are aware there is a decent AI solution on the market.

- Running a design workshop? Have [Miro](https://miro.com/aq/ps/assist/) or [Figjam](https://www.figma.com/figjam/ai/) synthesize your stickies for you.
- Reading a PDF? Have [Adobe](https://www.adobe.com/acrobat/complete-pdf-solution.html) or some alternative generate a summary.
- Conducting research? Start your search in [Perplexity](https://perplexity.ai/pro?referral_code=MWI6OCW4).

It’s likely these tools will come easy to you, but that doesn’t mean they are without their limits or friction. Take note of how quickly you were able to get up to speed, how well the experience blends into your daily habits, and how accessible they are when you aren’t thinking about them.

What makes for a sticky, useful, and usable AI experience?

## 6. Apply AI to irregular tasks

Once you’ve become more familiar with how AI could be incorporated into your workflow, start to find uses that don’t come as natural.

- Going for a promotion? Have AI conduct a mock interview for the job.
- Collecting links? Store them in a database in [Notion](https://www.notion.so/product/ai) or [Coda](https://coda.io/product/ai) and have the AI fetch information about the link in its own column.
- Analyzing data? Have [Julius](https://julius.ai/) or [Tableau](https://www.tableau.com/products/tableau-ai) conduct an analysis for you.

This might be tasks you don’t perform regularly, or more difficult tasks that you have built deep habits around solving.

Critically, AI might not help in all of these situations. You’ll start to find yourself asking, if AI really necessary here? You’ll get stuck. You might even get frustrated.

Force yourself to use AI for as many tasks as possible for several weeks. As a result, you’ll develop tenfold the empathy for users who are being bombarded by AI at every turn. Perhaps the products could be better designed–but you’ll also appreciate areas where product teams have assumed AI is a net good, when it fact it generates pain.

## 7. Explore advanced prompting techniques

At this point, you have a solid understanding for how the foundational models of AI work, a nuanced understanding of AiUX, and practical experience applying AI to actual jobs to be done.

Now, switch your thinking to how we build AI products. Start to explore prompt engineering.

There are many great resources available to you. Focus first on using [prompts created by others](https://docs.anthropic.com/en/prompt-library/library) to study their approach and the results they produce. Then, familiarize yourself with different [prompting techniques](https://www.perplexity.ai/page/advanced-prompt-engineering-ba-nVO4BvZiRwCdjio3g6Zq_Q) and begin to generate your own.

Play with adding [references](https://www.shapeof.ai/patterns/references) and [primary sources](https://www.shapeof.ai/patterns/primary-sources) to your prompts. How does this change the result?

Develop your own prompt library and start to use it in your regular habit of exploring different AI products.

## 8. Have the AI prompt for you

Just as AI can tell you which tokens it’s using, [AI has gotten very good at generating its own prompts](https://spectrum.ieee.org/prompt-engineering-is-dead).

Advanced techniques like RAG can still help you communicate your intent to AI. However, when it comes to crafting the actual input, AI might be better at this than you.

The reality is that this is happening behind the scenes whether we intend it to or not. The AI engine captures our input, translates it to tokens, and [regenerates the instructions in its own language](https://www.perplexity.ai/search/tell-me-how-ai-adjusts-my-prom-eeCKbXMpR0._X5d2r3BDug). (And some of these regenerations [can get really strange](https://www.oneusefulthing.org/p/embracing-weirdness-what-it-means)).

Understanding [how AI logically reworks your prompts](https://www.shapeof.ai/patterns/prompt-transparency) can help you better understand how the technology works, and how to help users get the best outcomes out of their own inputs.

So let off the gas a bit. Work _with_ AI. Set its direction, orchestrate its output, but let it do what it does best: speak in the language of computers and handle mundane tasks.

## Final thoughts

This final step should help you realize something important. The value of AI is not our control of it. Even when we think we are in control, we often aren’t.

Its inherent value is the ways it can teach us about information, make us more self aware about what and how we share, and let us build experiences that center humans within technology.

The goal of learning AI is not to give away your job. Rather, it’s to understand our relationship to technology in new ways. When we let algorithms dictate what we see, and subconsciously control our behavior, we trade our autonomy for convenience.

AI let’s us communicate our intent directly to the computer.

We work with the clay, we shape it with the wheel, but we are ultimately in charge.

With AI, we can orchestrate our own digital experience.

But first, we have to design it.

---

## 🎉 Bonus: Share your experience

Here’s the reality: you’re going to get stuck.

You will reach moments of frustration and confusion. The AI won’t respond the way you want it to. [It will hallucinate](https://www.ibm.com/topics/ai-hallucinations). [It will get lazy](https://www.inc.com/ben-sherry/chatgpt-is-showing-signs-of-laziness-openai-says-ai-might-need-a-fix.html).

You will run into product experiences that don’t make sense.

You will get tired of AI.

All of this is ok and part of the learning experience. My best advice? Share it!

Learning AI is like trying to shoot an arrow at a moving target. By writing or talking about your experiences, you’ll benefit others, and it will help you work through the friction.

It will make you more aware of your experience. It will make you a better designer.

Share your experience so others can learn with you–and tag me! I would love to follow your journey.

❋ Emily

---
*Originally published at [https://emilycampbell.co/writing/shaping-the-clay-of-AI](https://emilycampbell.co/writing/shaping-the-clay-of-AI) on 2024-04-25.*


# Exploring the spectrum of “Needfulness” in AI Products

Not every AI product has a clear commercial purpose, but it should solve for some compelling user interest. We know this is the case when people have full information about what the experience is, what data it collects, and how it uses that data, and then they still choose to opt-in without the need of dark patterns and incentives.

[!callout]
Need a quick reference? [↓ Download the article guide](https://drive.google.com/file/d/1asoyaq_WOK1boHKYD-Zs0zhWx-_BXbnI/view?usp=sharing), covering the range of needfulness in AI products and considerations when designing for each.

In the race to market, we’re already seeing a proliferation of flat out *bad* AI experiences hit consumers. A consistent thread I see connecting many of these are questions along the lines of, “who ever thought we needed this?”

Discord [shut down its chatbot, Clyde](https://www.theverge.com/2023/11/17/23965185/discord-is-shutting-down-its-ai-chatbot-clyde) last fall with no explanation, though user reviews hint at a [horrendous experience](https://thetaxcollectorman.medium.com/discords-new-ai-chatbot-is-a-useless-miserable-nightmare-c7385095ec89) that made the overall product worse. Google [removed the ability for it’s AI, Gemini to create images](https://www.nytimes.com/2024/02/22/technology/google-gemini-german-uniforms.html) after users found its “historically accurate” photos were anything but.

This isn’t limited to chatbots. LinkedIn’s Generative Articles are positioned as a way to let AI-crafted prompts drive human discussion, but in practice they are incentivizing users to [game the system](https://www.linkedin.com/posts/danwiner_im-no-longer-a-top-voice-of-user-experience-activity-7143219535091871744-T9Ih), or to contribute [low-value responses](https://medium.com/illumination/what-went-wrong-with-linkedin-ai-collaborative-articles-e46be135b7e6) to [poorly-framed prompts](https://www.linkedin.com/posts/thinkingslow_linkedins-collaborative-articles-are-an-activity-7120384835377840129-thM6/). Don’t get me started on their text generation tools. *Who needs this?*

![](https://emilycampbell.co/images/blog/exploring-the-spectrum-of-needfulness-in-ai-products-img-2.webp)

And this effect has spilled over into non-AI products. [Philipp Temmel](https://www.linkedin.com/in/philipp-temmel-753bb3146/) had a [great piece recently in Creatively](https://creativerly.com/what-happened-to-product-hunt/) about the ways this surge of unnecessary and poorly crafted products has degraded platforms like Product Hunt.

It’s the natural next step to what [Cory Doctorow](https://craphound.com/) coined “the [enshitification](https://doctorow.medium.com/social-quitting-1ce85b67b456) of the web.”

> When switching costs are high, services can be changed in ways that you dislike without losing your business. The higher the switching costs, the more a company can abuse you, because it knows that as bad as they’ve made things for you, you’d have to endure worse if you left.

> Cory Doctorow [https://doctorow.medium.com/social-quitting-1ce85b67b456](https://doctorow.medium.com/social-quitting-1ce85b67b456)

*The **AI-shitification** of the web is real.*

## Why is this happening?

Adding AI to a product strategy drives big increases in valuation, despite the fact that, [many of the companies receiving these big rounds from VCs have yet to deliver any revenue](https://pitchbook.com/news/articles/generative-ai-vc-huge-valuations-small-revenue).

Even for established players, VCs are offering [4x, 6x or 10x valuation jumps](https://news.crunchbase.com/ai/valuations-venture-not-slowing-2024/) year over year (or less). Similar trends are playing out for public companies embracing AI.

- Companies are incentivized to add AI…
- …so that they can say that have AI…
- …so they can collect the AI-sized check.

However, these companies are not yet accountable to ensure that what they have implemented is viable or useful to users.

[!callout]
The hype multiplier is real.*** 🤑

Eventually this will come out in the wash. Products that don’t provide real value will lose to those that do, and the market will consolidate.

But there will be a ton of debris leftover: UX debt, zombie legacy features, and crummy AI-driven (or written) content.

This will harm consumers. This will make the overall experience of the web worse, even as the wave recedes. We should be taking this seriously.

## So, does it matter that an AI product serves a critical purpose?

The short answer is, no.

AI isn’t a monolith, and it’s not a solution; It’s a tool. Its use should be determined by the need or opportunity, and not the other way around.

There are certainly use cases where AI serves an essential need. It can help people do things they couldn’t do themselves, or help them save significant effort to get those jobs done.

On the other hand, some needs are less obvious, such as the need for fun, to spark curiosity, connect with others, or even waste time. Tools don’t need to have a specific commercial value (e.g. monetizable) to provide a good experience. They do need to provide enough of something that users are compelled to use it by choice.

I’m concerned with the uses that aren’t explicitly needful *and* that exist to serve the business *at the expense of users.* I would put LinkedIn’s functionality pretty squarely in this category.

So there’s a spectrum of use cases. As designers, product people, and researchers, we should be mindful of which we are working on at any moment, and adjust our approach to creating constraints, nudges, and incentives accordingly to ensure the experience centers on users and their needs.

## Four categories of needfulness

![](https://emilycampbell.co/images/blog/exploring-the-spectrum-of-needfulness-in-ai-products-img-3.webp)
### 1. Trap doors

When a product appears to exist primarily to serve some business need at the expense of users, these are trap doors. They may be connected to some incentive or feature that has the veneer of being user-centered, and some people may even find benefit from them. However, their central purpose is business need over user need.

- **AI-prompted articles** that serve as a clever SEO strategy, and are likely are helping the company train their eventual LLM, are not user-centered, even if users get “top contributor” badges for playing. People need to know the full experience they are opting into to truly evaluate – gamified incentives affect free will.
- **Meta’s chatbot**, which blatantly lays out in its terms that it will scrape user conversations for training data without any ability to opt-out, are not user centered, even if someone has an entertaining conversation with the AI (🤔 in fact, engaging conversations probably lead users to give *more* personal or revealing data to the AI).
- Forcing **Search Generative Experiences (SGE)** on [users who didn’t opt into them](https://arstechnica.com/gadgets/2024/03/google-search-will-start-automatically-showing-a-chatbot-to-some-users/), is not user centered, even if some users do get to a good answer in the result. The justification by the most popular search engine in the world that a broad rollout is justified to “get feedback and learn how a more general population will find this technology helpful” is not sufficient, especially when it appears the results are clearly [not ready](https://getstat.com/blog/sge-is-not-the-answer/) for [prime time](https://www.linkedin.com/posts/emmiecampbell_ai-kids-aiux-activity-7171860557497581568-fdwA).

![](https://emilycampbell.co/images/blog/exploring-the-spectrum-of-needfulness-in-ai-products-img-4.webp)
*Note: I recognize that unfortunately designers often don’t have the ability to choose what they work on. If you are unable to push back on dark patterns and incentives that are mandated by the business, you can still compensate with [strong caveats](https://www.shapeof.ai/patterns/caveats) and [identifiers](https://www.shapeof.ai/pattern-types/identifiers) so users have full knowledge of what they are interacting with.

### 2. Vitamins

AI products or features can be simple, functional, or fun without serving a clear user need. Unlike “painkillers” (borrowing from the [common product strategy framework](https://www.entrepreneur.com/starting-a-business/is-your-product-a-vitamin-or-painkiller/230736)), these are nice-to-haves, and therefore have a harder path to viability.

As with Trap Doors, the lack of viability means that businesses need to find other ways to derive value from them in order to justify the investment. Because Vitamins are often introduced in a marketing funnel or as a side-car experience tied to the brand, their value derives from web traffic, lower CACs, or increased conversion rates, making them less susceptible to dark patterns.

- Several of [**Ramp’s experiments**](https://ramp.com/mission-statement-generator) incorporate AI, such as their [mission statement generator](https://ramp.com/mission-statement-generator). Each is connected to useful lessons about business operations, and while they are used for marketing purposes, their simplicity and their ability to function without personal information make them fairly benign.

- [**GoDaddy’s AI Domain search**](https://www.godaddy.com/domainsearch/find?domainToCheck=Peony#) and similar products essentially remove the step of manually searching for domains in your focus area. That’s a fairly small boost in efficiency, but it also helps educate users on how to find available domains by combining key words, and increases the likelihood of conversion.
![](https://emilycampbell.co/images/blog/exploring-the-spectrum-of-needfulness-in-ai-products-img-5.webp)
- **Automatic meme generators** appear to be a new cottage industry, with fun tools like [MemeCam](https://www.memecam.io/) and [MemeDaddy](https://memedaddy.ai/) popping up. I have no idea when I would need a tool that creates silly captions for random images, but I could totally see a company developing a fun tool like this to drive visitation.

### 3. Augmenters

One of AI’s superpowers is its ability to make mundane tasks feel lighter or easier. That’s where augmenters come in. These products clearly save users time or money, and they are willing to pay for the service.

- [**Perplexity.ai**](http://perplexity.ai/) combines AI-generated summaries of search questions, combined with the resources that back up its response. Users can limit resources to more credible types, decreasing the effort it would take to conduct initial reviews.
- [**Github’s copilot**](http://github.com/features/copilot) has already demonstrated a sizable ROI through its ability to generate sample code and answer questions without requiring the developer to leave the IDE.
- [**Hypotenuse.ai**](http://hypotenuse.ai/) combines a template library, easy-to-build prompts and other content tools with an automated SEO analyzer, combining multiple operations into one and making them faster to manage.

### 4. Access enablers

These uses of AI helps remove barriers and exclusionary experiences by compensating for capabilities that are [inaccessible due to permanent, temporary, or situational circumstances](https://inclusive.microsoft.design/tools-and-activities/Inclusive101Guidebook.pdf). They serve a critical need by helping users complete tasks they are otherwise unable to perform, often in a multi-modal way.

- **Zoom’s auto-generated captions** allow all participants in a meeting to follow along regardless of their auditory capabilities at the moment.
- **Automatic image descriptions** allow users of [Be My Eyes](https://www.bemyeyes.com/blog/introducing-be-my-eyes-virtual-volunteer) to upload an image and have the AI describe it, which can be read back using text-to-voice. Microsoft and Google have added similar capabilities to their bot development framework to make it simple for developers to auto-generated alt text when images are shared.
- [**AI-driven translation**](https://blog.google/products/translate/new-features-make-translate-more-accessible-for-its-1-billion-users/) tools allow people to communicate more fluidly across languages, a fun addition to travel and critical for people who need to communicate accurately and urgently outside of their native tongue.

## Considerations for different levels of needfulness

I’ve assembled a list of specific considerations and success criteria that you can use to evaluate your design of different types of AI Products.

These are intended to be a starting point, not exhaustive. Use them to prompt critical thinking and conversations with your team about building intentional experiences.

![](https://emilycampbell.co/images/blog/exploring-the-spectrum-of-needfulness-in-ai-products-img-6.webp)

---
*Originally published at [https://emilycampbell.co/writing/exploring-the-spectrum-of-needfulness-in-ai-products](https://emilycampbell.co/writing/exploring-the-spectrum-of-needfulness-in-ai-products) on 2024-03-28.*


# My emerging heuristics for assessing AI Design

*How do we define “good” AI Design?*

This question has been swirling in my head every time I interact with a new AI-powered product or feature.

- **It has to be usable**, both in its raw form and within the context of whatever product or wrapper you’re accessing it from. Thirty years in, [Jakob Nielsen](https://www.linkedin.com/in/jakobnielsenphd/)‘s [Ten Usability heuristics](https://www.nngroup.com/articles/ten-usability-heuristics/) remain as useful and relevant as ever.
- **It has to be familiar, fluid, and understandable**, so we can “converse” with it or direct it without requiring significant [code-switching.](https://en.wikipedia.org/wiki/Code-switching) [Erika Hall](https://www.linkedin.com/in/erikahall/)‘s [Conversational Design](https://abookapart.com/products/conversational-design) guides us to consider how we can design programs that *speak human* fluently through universal principles.
- **It has to be ethical and trustworthy**, so we can responsibly and securely use it in different contexts. I’ve been following [Yaddy Arroyo](https://www.linkedin.com/in/yaddy/)‘s [explorations into AI Ethics](https://www.linkedin.com/posts/yaddy_%3F%3F%3F%3F%3F%3F-%3F%3F-%3F%3F%3F-%3F%3F%3F%3F%3F%3F%3F-%3F%3F%3F%3F%3F%3F%3F-activity-7171555000547041281-UP99), which offer a blueprint for evaluating how ethics shape the user experience along their entire journey.

Each of these frameworks covers an aspect of AI design, but there’s a missing piece for me: Our relationship to AI has changed.

### Until recently, our relationship with AI was one-directional.

AI could predict what information we were searching for, but users couldn’t direct it back.

Now, that’s changed. Rapid advancements in Natural Language Processing have fundamentally reconfigured the landscape of AI products. Users can now personalize the AI’s instructions, provide contextual feedback in real time, and give references and specific guidelines that describe the outcome they are seeking.

But there’s a catch. The more autonomy that users have to direct computers to their personal whim, the better they will expect computers to understand them and anticipate their needs.

**Perception matters.**

As users come to expect AI to adapt to their needs, their [inflated expectations can result in a poor user experience](http://library.usc.edu.ph/ACM/CHI2019/1proc/paper411.pdf) if they aren’t met.

But this isn’t just about perception. [Users have more trust in systems that they can direct](https://dl.acm.org/doi/pdf/10.1145/3584931.3606997), even if the system makes a mistake. The bi-directionality of working with computers as digital assistants and agents creates a new paradigm for HCI.

This will only continue. Gradually, computers are becoming more autonomous themselves, and more capable of sensing changes in our contexts, emotional states, unconscious needs, etc.

So, how do we evaluate the UX of AI as it grows more adaptive and integrated into our everyday lives?

## Taking a page from Adaptive Systems Theory

Adaptive Systems offer hints for how we can adjust our understanding of User Experience to this new paradigm.

Some of the ways that well-designed AI experiences experiences mimic* these principles include:

- **Feedback loops:** these experiences use mechanisms to receive input from users and adjust accordingly.
- **Flexibility and Personalization:** they give users control to tailor or tune the experience to meet their needs.
- **Anticipation of needs:** they use data to proactively anticipate user needs and improve, “learning” and modifying over time.
- **Resilience:** they treat errors as inputs, and not fail states, since users can redirect the system to get a better outcome.
- **Context-awareness:** they can react and adapt their behavior to different situations and constraints.

*(I say "mimic" because until we reach true AGI, AI-driven products will always require some sort of user interactivity to adjust and personalize its response systems)*

From a UX perspective, we can use these principles to paint a picture of AI interaction design. Like any two-way interaction, AI design relies on loops of information, powered by capabilities that ensure a program has sufficient input to adapt to the user’s needs as the conversation progresses:

- I need to have the control to craft my input,
- I need to receive enough information back to improve my input, and
- I need the output to be trustworthy and accurate so I am willing to re-engage.

## Evaluating what “good” AI looks like?
![](https://emilycampbell.co/images/blog/my-emerging-heuristics-for-assessing-ai-design-img-2.webp)

Blending all of this together, I’ve developed my own set of heuristics that I’ve been using to evaluate AI products, particularly products that include generative capabilities and natural language processing.

These are emerging best practices as I see them, not specific guidelines.

When these are in place, information flows intentionally. It feeds better inputs, better outputs, faster adaptability, and higher trust.

I go through each in detail below. Here’s the topline list:

## Heuristics for AI design

- **Purposeful and Needful**: The AI solves for a real and significant need in a meaningful way, and makes sense within its surrounding context.
- **Input Clarity and Ease**: It is intuitive and easy for users to build prompts that result in accurate, relevant, and high quality results.
- **Result Quality and Context**: Outputs are clear, accurate, and relevant, and supplemented with additional information to enhance user comprehension and provide helpful context.
- **Customization and Tunability**: Users can fine-tune their inputs to easily generate outputs that match their specific needs and expectations.
- **Branching and Recall**: The AI maintains conversational context throughout or across interactions, allowing users to explore different paths and easily return to the main thread.
- **User Autonomy and Control**: Users maintain control over the AI through mechanisms that let them guide, direct, and control the interaction.
- **Logical Transparency**: The AI clearly communicates its decision-making processes, improving user comprehension and trust.
- **Continuous Learning**: The AI continuously improves, learning from user feedback and data to enhance its functionality.
- **Ethical Integrity and Trustworthiness**: The AI adheres to ethical standards, minimizes bias, protects privacy, and ensures transparency to foster user trust.
- **Identification and Honesty**: Users can distinguish AI inputs and outputs from human-generated content. The AI is honest and transparent about its capabilities and constraints.

## #1 Purposeful and Needful

*The AI solves for a real and significant need in a meaningful way, and makes sense within its surrounding context.*

We are in the “[put a bird on it](https://www.youtube.com/watch?v=GNpIOlDhigw)” moment in the AI hype cycle. It’s a race to the starting line. For every product implementing a useful feature that makes work or life easier through AI, there are 3 more examples of meaningless clutter pushed on us to show that the company can DO AI.

The list of problems that AI can be applied to will continue to grow as the technology evolves. The importance of introducing this technology only when it’s purposeful won’t change.

![](https://emilycampbell.co/images/blog/my-emerging-heuristics-for-assessing-ai-design-img-3.webp)
*LinkedIn’s “Rewrite with AI” capability exists as a standalone feature with no clear purpose to the average user. Also it’s absolutely hilarious that this is how their AI re-wrote my comment.*

**Sample Assessment Criteria:**

- Could this task be successfully completed just as fast without AI?
- Are the outcomes significantly better?
- Does this exist to solve a human need or a business need first?
- Are the incentives behind the implementation of this AI clearly centered on user needs?

## #2 Input Ease and Effectiveness

*The AI allows users to easily and intuitively craft a prompt that is likely to give them a useful response.*

The giant input box that asks “what do you want to build today?” has become a gimmick in AI platforms. After a few interactions, users quickly realize they will need longer and more specific prompts to return an output that meets their needs. Or, they disengage.

There’s a gap between the simplicity of this common UI pattern and the required complication of an effective prompt. *(See [Microsoft’s training video on advanced prompting](https://www.youtube.com/watch?v=geQy3mkKx-0) as an example)*

Great experiences carry the load for the user by mapping the input type to the context (not everything needs to be open text), allowing references or examples to reduce how much information needs to be manually inputed, and using [suggestions](https://www.shapeof.ai/patterns/icebreakers-suggestions), [templates](https://www.shapeof.ai/patterns/template), and [other wayfinding devices](https://www.shapeof.ai/pattern-types/wayfinders) to help the user craft an input that works.

[Writer](https://www.linkedin.com/company/getwriter/) has a fantastic sample prompt library that helps users easily craft advanced prompts. They also benefit from seeing examples of what good prompts look like.

Great AI products that don’t center on the generative input make it easy to “call” AI ([GitHub](https://www.linkedin.com/company/github/) co-pilot or [Figma](https://www.linkedin.com/company/figma/)‘s Figjam assistant), craft inline prompts or inputs ([Klu.ai](https://www.linkedin.com/company/klu-ai/)‘s playground or [Notion](https://www.linkedin.com/company/notionhq/)‘s AI-prompt column-type), and re-generate the request (Midjourney’s image regeneration button).

![](https://emilycampbell.co/images/blog/my-emerging-heuristics-for-assessing-ai-design-img-4.webp)
*Writer.com offers prompt templates to make it easy to build complicated inputs that work. On top of that, users benefit from seeing what a good prompt looks like, helping them learn.*

**Sample Assessment Criteria:**

- How many attempts would a user need to take to get an output that feels right
- What guidance does the system offer to help the user?
- How does this vary depending on where the user is in their customer lifecycle, their capabilities, or their preferences?

## #3 Result Quality and Context

*Outputs are clear, accurate, and relevant, and supplemented with additional information to enhance user comprehension and provide helpful context.*

If a user can’t asses whether the AI provided an accurate and reliable result, it’s not a useful or usable system.

Quality results are partially a result of great inputs, which is why *Input Ease and Effectiveness* matters. However the choice of model, programming of the model, underlying prompts etc also shape the outcome. The difference is, a user can’t see or control these.

A user also can’t control the data that a response is based on, which is why showing the results or letting a user trace back to them through other means is critical to helping them assess the accuracy and usefulness of the output.


![](https://emilycampbell.co/images/blog/my-emerging-heuristics-for-assessing-ai-design-img-5.webp)
*Google’s AI powered search lists mental health disorders as the first possible explanation for “Why are my friends mean to me.” This could be harmful, but by showing the sources, the user can identify that this poor result comes from a Quora article.*

![](https://emilycampbell.co/images/blog/my-emerging-heuristics-for-assessing-ai-design-img-6.webp)
*Slack shows the importance in a business context of showing the work behind its results.*

**Sample Assessment Criteria:**

- How often does the system exclude information that should be included? How often does it hallucinate or give false information?
- Do the results remain accurate throughout the conversation or do they degrade?
- Are sources and other [footprints](https://www.shapeof.ai/patterns/footprints) provided to help a user identify the source of information?

## #4 Customization and Tunability

*Users can fine-tune their inputs to easily generate outputs that match their specific needs and expectations.*

AI that personalizes its results to the user is central to adaptive UX. Particularly in generative settings, users will expect that the AI produce content that matches their voice and tone, technical level, and overall brand.

Overtime, AI will learn to adapt these personalized user preferences and characteristics automatically. For now, good AiUX allows users to explicitly set their own parameters and tweak their settings to ensure the AI’s output meets their needs.

These personalized nudges can be set at the conversational or thread level, or targeting a specific prompt:

- At the overall conversational level, a user may choose to constrain results to a specific modality, to capture a specific tone of voice in a generative output, or define which model to interact with.
- At the prompt level, a user may direct the AI to set a specific length for its response, capture certain tokens, or weigh one token to attribute over another.

![](https://emilycampbell.co/images/blog/my-emerging-heuristics-for-assessing-ai-design-img-7.webp)
*Jasper.ai makes it easy for users to set parameters up front and tune them later.*

**Sample Assessment Criteria:**

- Can the AI consistently reproduce outputs that follow specific characteristics, and then maintain those characteristics throughout a conversation?
- Does the AI system provide clear guidance on how to adjust settings for optimal results?
- If a user intervenes to direct the AI on a specific way to tune its output, how responsive is the AI to that direction?

## #5 Branching and Recall

*The AI maintains conversational context throughout or across interactions, allowing users to explore different paths and easily return to the main thread.*

Natural conversation is not linear, and we should not expect that a user’s interaction with AI will be either. AI can support the affordances of dialogue by making it easy for a user to follow a thread and then return to the main branch of a conversation.

This is not limited to AI Conversations. Giving users the ability to reference other threads leads to a more fluid and human interaction. For example Midjourney allows users to reference previous results when crafting new ones to inform the AI of their intent. ChatGPT’s introduction of GPTs that can be mentioned in other conversations allows the user to direct the AI without leaving the context of their main thread.

![](aiheuristics-6%201.png)
*ChatGPT allows users to mention GPTs in threads, giving the user the ability to perform multiple interactions within a single thread without losing its core context.*

**Sample Assessment Criteria:**

- Does the AI effectively maintain context over interactions?
- How easily can users navigate through complicated conversation paths and find their way back to previous comments?
- How easy is it to reference or find older thoughts or comments?
- Is the AI constrained to a single thread or interaction or can it reference information in other threads to remix with?

## #6 User Autonomy and Control

*Users maintain control over the AI through mechanisms that let them guide, direct, and control the interaction.*

[HAL 9000 is an anti-pattern](https://www.youtube.com/watch?v=Mme2Aya_6Bc).

Good AI is human centered: solving real human needs in a personalized way. As the technology advances, it is critical that users retain the ability to explicitly direct the AI.

At the surface level, this could be as simple as “stop and go” controls that allow users to abort generation of a result if its going off the rails. Implicitly, users will come to expect advanced Natural Language Processing in their conversational interfaces, giving them the ability to guide the AI to a specific output without having to learn the speak the computer’s language.

![](https://emilycampbell.co/images/blog/my-emerging-heuristics-for-assessing-ai-design-img-8.webp)
*Sometimes ChatGPT will stop following an explicit instruction with no explanation. The platform gives users the ability to stop a generation, so you can catch the error and re-run the request.*

**Sample Assessment Criteria:**

- Can the user start or restart a Generative AI result if it’s not meeting the user’s need mid-prompt?
- Can the user direct the AI through a list of instructions (visit this link, review my format) before beginning a response? How frequently does the AI ignore these instructions?
- Does the AI act autonomously or follow the user’s lead? Does the AI get [lazy](https://www.inc.com/ben-sherry/chatgpt-is-showing-signs-of-laziness-openai-says-ai-might-need-a-fix.html)?
- Can the user exclude specific tokens, sources, or other inputs from the final product?

## #7 Logical Transparency

*The AI clearly communicates its decision-making processes, improving user comprehension and trust.*

Transparency is critical to building user trust. An AI that can articulate “how” and “why” it generated some result empowers a user, making the technology more approachable, reliable, and configurable.

The clearest issue with transparency relates to the opaque training data of popular Large Language Models (see the [NYTimes lawsuit against OpenAI](https://www.nytimes.com/2023/12/27/business/media/new-york-times-open-ai-microsoft-lawsuit.html?unlocked_article_code=1.dE0.PY8t.bODq98_bzB48&smid=url-share), or the inability of the OpenAI CTO to specify the [training data behind their new video generation model](https://youtu.be/mAUpxN-EIgU?si=EdOJkBAfkGt2G5r5&t=263)).

This extends beyond elucidating the logic behind a specific result. Other considerations include:

- Training data and ethics
- Efforts to combat bias in the modal
- Limitations to the model’s results
- Advice for how to improve the results

Within smaller and proprietary models, this is more an issue of being able to work backwards from a result to improve the model itself. Understanding how AI came up with a result gives us critical inputs to improve the results.

![](https://emilycampbell.co/images/blog/my-emerging-heuristics-for-assessing-ai-design-img-9.webp)
*Currently consumer-facing AI products rely on basic caveats instead of providing specific details about their training data and specific likelihood of inaccuracies.*

**Sample Assessment Criteria:**

- Can the AI articulate how it arrived at a specific answer?
- If it gets information wrong, can it explain why or does it only correct its result?
- Can the AI give the user instructions for how to get better results?

## #8 Continuous Learning

*The AI continuously improves, learning from user feedback and data to enhance its functionality.*

Adaptive systems can learn and adjust to meet an infinite number of contextual clues and user needs. Without the ability to learn from each of these adjustments, AI will rely on user input to improve its results.

Direct input from users takes work and is prone to error. The faster and more reliably AI can learn, the more personalized and human-centered each subsequent interaction can become, and the faster AI can deliver more agentive capabilities.

In addition to learning the preferences of its users, AI can support more scalable systems that improve over time by learning more efficient or effective ways to produce the results users want.

![](https://emilycampbell.co/images/blog/my-emerging-heuristics-for-assessing-ai-design-img-10.webp)
*Nest thermostats offer a timeless example of Adaptive AI demonstrating continuous learning in a human-centered way. The more the user interacts with the thermostat up front, the faster the technology learns and anticipates user preferences.*

**Sample Assessment Criteria:**

- Does the AI reflect a growing understanding of the user over time?
- Can the AI recall the user’s adjustments and improve its results?
- Do the AI’s learnings persist beyond single interactions?
- Does AI anticipate emerging user needs and preferences before the user is self-aware of them?
- Does the AI begin to proactively anticipate and avoid error states?

## #9 Ethical Integrity and Trustworthiness

*The AI adheres to ethical standards, minimizes bias, protects privacy, and ensures transparency to foster user trust.*

As interfaces change from being screen-based to conversational, our understanding of [minimalist design and aesthetics](https://www.nngroup.com/articles/aesthetic-minimalist-design/) must evolve. Ethics have traditionally been considered an [aesthetic aspect of design](https://link.springer.com/article/10.1007/s00146-021-01279-w), as the reflection of how something looks and feels should reveal the functional incentives, biases, and constraints of its builder.

Ethical integrity goes beyond the model and implementation itself. Other considerations for the ethics of the technology include:

- how the model acquires new data to train on,
- clarity in its documentation,
- multi-modality of its outputs to support users with different physical or cognitive abilities

and more.

Trustworthiness is both a reflection of the transparency of the model, and of the reliability of its results. Proactive error avoidance, continuous learning of user preferences, and logical guidance to improve inputs and outputs feed into user trust.

![](https://emilycampbell.co/images/blog/my-emerging-heuristics-for-assessing-ai-design-img-11.webp)
*Large Language Models bury the ability to opt out of letting the model train on your data deep in user preferences. Meta AI doesn’t allow users to opt out.*

**Sample assessment criteria:**

- Is the product transparent about the underlying model it uses and the model’s limitations?
- Can users control how their data is used and accessed?
- Is the model’s owner transparent about any underlying biases and the actions they have taken to adjust for them in results?
- Does the AI support multi-modal inputs and outputs ensure it is accessible to users with different capabilities and circumstances

## #10 Identification and Honesty

*Users can distinguish AI inputs and outputs from human-generated content. The AI is honest and transparent about its capabilities and constraints.*

Imagine a conversation where someone wasn’t who they said they were, or were intentionally concealing their knowledge about the topic of discussion.

For AI experiences to be human-centered, users must know at all times when they are interacting with an AI or with AI-generated content, and what the limitations of the AI are.

Furthermore, AI should be honest about what instructions it is following and when it is acting autonomously. Individual users may choose to allow AI to interact with them in a less bounded way. In this case, defaults matter. Granting more freedom to the AI should be a preferences that someone can opt *into* so no one is caught off guard, unable to find the boundaries between person and machine.
![](https://emilycampbell.co/images/blog/my-emerging-heuristics-for-assessing-ai-design-img-12.webp)
*iA writer supports the co-creative nature of Generative AI while keeping the user informed and centered. Text generated by the machine is clearly distinguishable from the user’s own voice.*

---
*Originally published at [https://emilycampbell.co/writing/my-emerging-heuristics-for-assessing-ai-design](https://emilycampbell.co/writing/my-emerging-heuristics-for-assessing-ai-design) on 2024-03-17.*


# Designing with stories

One of the best lessons I learned about design and product leadership came from the guys who made Southpark (yup, you heard me right). Two words: “But” and “Therefore.”

These two words give shape to stories. **BUT** represents conflict or opportunity. **THEREFORE** pulls the story forward. They stir emotion. They get people invested in the outcome.

Why is this one of the best lessons I’ve applied to my craft?

Because our ability to influence others starts with stories:

- Here’s why investing in this feature will lead to growth
- Here’s how a small usability improvement could have a big impact
- Here’s what I learned after speaking with 10 users

And yet these often fall flat.

Too often, we fail to build a narrative. We aren’t giving shape to our stories. Others can’t see and feel what we do.

Take this example: Your team learns that users are frustrated that they can’t export their data. Executive leadership is concerned with people doing tasks off platform. You want to convince them that allowing them to do this will INCREASE engagement. You tell them, if we build this, then customers will be happier, so they’ll be more likely to keep using our product. “And then and then and then” ... The executive doesn’t see it.

What if it was presented like this:

> [!note]
> 💝 People like that we aggregate their data. 83% of people we spoke with said it was a primary reason they bought us
>
> 😫 **BUT** they are frustrated that they can’t do more to manipulate it
>
> 💸 **THEREFORE** it undermines the value of the rest of the platform to them, even though we have made an intentional decision not to invest in this functionality. This has caused some customers to churn to a competitor
>
> 💡 **BUT** they tell us that if they could export the data regularly, it would mean our product is the source of truth, even if it isn’t the source of data manipulation
>
> 🔥 **THEREFORE** they wouldn’t feel it was an urgent to replace us with a competitor that had these types of data tools

*[I’ve actually worked on this problem]*

We have a responsibility to see how problems connect, and how our solutions can address these problems together. When design can add value upstream (customers having more confidence in our features when they are sold) and downstream (decreasing the likelihood of churn), then our value to the business is clear.

You’re going to find your own words to express this, but when you’re getting started, right it out using this format. BUT, THEREFORE. Show the shape of your story and you’ll be more successful helping others see it too.

---
*Originally published at [https://emilycampbell.co/writing/designing-with-stories](https://emilycampbell.co/writing/designing-with-stories) on 2024-03-08.*


# The Pace Layers of SaaS Organizations

Hot take: any issue you find in a user experience points to a problem somewhere deeper in the company. Great cultures produce great products. Poor cultures, well…

*Success has many parents, failure is an orphan.*

I’m sure you are already thinking of a few examples.

The problem is people are most likely to react to what they see. Everyone has an opinion about the interface, CEOs dictate copy, we blame the design when a user runs into a problem. But…

- …maybe the design was based on the wrong assumptions of how the product is used by the customer, but the designer never had the opportunity to look
- …maybe those assumptions exist because we prioritized a business metric over listening to customers and serving their needs first
- …maybe someone had that realization but they weren’t empowered to do anything about it
- …maybe they were actively blocked
- …maybe internal politics incentivized them to stay quiet

Sound familiar? What can we do about it?

## Pace layering as a diagnostic

[The Pace Layering model](https://jods.mitpress.mit.edu/pub/issue3-brand/release/2) helps me make sense of this. The idea is that a system has multiple layers. Deep-rooted layers affect everything on top of them, evolve slower, and tend to be less visible (think “culture”). Surface layers are more reactive to what’s beneath them, and change faster (think “fashion and trends”).

I could list 10 things designers could be doing differently to address this, but let’s dig deeper. CEOs, operational executives, Heads of HR, People leaders — this is for you. Your product is a beacon of the health of your organization and processes.

If the system is producing poor results, fix the system. When a canary dies in a coal mine, we don’t blame the bird.

I know this is heavy, but let’s have more conversations like this. Companies are systems. Systems are a reflection of decisions.

Design systems with intent. Or don’t, but the results speak for themselves.

---
*Originally published at [https://emilycampbell.co/writing/the-pace-layers-of-saas-organizations](https://emilycampbell.co/writing/the-pace-layers-of-saas-organizations) on 2024-03-04.*


# A new pattern language for a new paradigm shift

**BLUF:** Product development in AI is accelerating faster than we can keep up. This is already leading to fragmented experiences and feature bloat. Design is more important than ever, but how can we navigate these changes at the pace of the market? We need to combine forces, and re-think common interface and interaction patterns for this new medium. I’ve started this process, but I need your help.

Today I’m launching [ShapeOf.AI](http://shapeof.ai/)– a documentation library of the emerging and adapting patterns already showing up in products using AI. What’s missing? What needs to change? What are you seeing that I’m not?

Please join me on the journey. This moment is an opportunity to reset our approach to technology, and to redesign design as a discipline. Hop on. ❋

## Everything everywhere all at once

AI feels like it’s everywhere, and it’s accelerating fast.

As features get baked into products, and new products appear en masse, the window is closing to take a thoughtful and collective approach to evaluate what interaction design looks like in this new medium.

What parts of the old paradigm make sense. What needs to change? What needs to be invented?

**Design is more important than ever.** If we don’t take a stab at walking into this together, we’ll find ourselves playing catchup.

Honestly, we already are.

We’re in a race to the starting line. Companies are prioritizing “doing AI,” but struggling to gracefully integrate it into their current products. It’s not clear if they are taking the time to assess how a feature should work, why it should work that way, and if it should even exist.

## A new interaction paradigm

Things aren’t all lost. Digital interactivity has evolved before.

- In the mid 20th century, people went from being the computers to training the computers.
- In the latter half of the century, we moved from training computers to interacting with them through GUIs. We no longer had to memorize code to get a computer to do what we wanted it to.
- In the late naughts, digital experiences went from being static to dynamic through algorithmic news feeds and cross-platforms integrations. We no longer could predict exactly what content someone would see at any given time within the same context.
- And now we are in a world where we are back to programming computers to do what we want them to do, but the list of possibilities has multiplied by a massive factor.

In each of these moments, many of our old patterns didn’t make sense any more. Some made things worse. Some required changes in patterns upstream: to how we organize, learn, and work together.

This time feels different though. It’s happening so fast! And it’s happening right under our noses. Here’s how I know…

## The patterns of AiUX

*[Side note, I’m going to use AiUX to refer to patterns and interactions that are specific to the AI medium]*

I’ve spent the last month cataloging emerging patterns in AI products. After wading through 40 or so different “solutions” on the market, there’s one thing I can say for sure.

Things are a mess.

Products have lots of cool stuff in them. But the stuff is not consistently aligned to a journey. And the stuff differs from experience to experience. This results in fractured products and confusing interfaces.

…But I think there might be something worse happening.

**Our limited understanding of AiUX patterns is restricting our ability to imagine and create products that are different from what’s already available. We are limiting our ability to innovate our experiences at the pace of change to the business and computational models powering the products themselves.**

As AI becomes increasingly integrated into digital products and services, the design of AiUX interactions will dictate the success and user adoption of these products.

However, if we fail to define and adopt new AiUX interaction patterns, relying instead on patterns established under heuristics for an old interaction model, we may be limiting the scope of innovation. This approach confines designers to a commodified box of existing solutions, and hinders the exploration of novel and potentially more effective interactions.

> Our limited understanding of these emergent UX patterns is limiting our ability to think beyond the commodified patterns (and therefore products) currently on the market.

It’s like trying to build a rocketship with materials developed to build a car. At some point, they are going to break.

## Let’s start here: What is the journey of AiUX?
![](aiheuristics-1%201.webp)
*A rough mental model for the journey of AiUX interaction. Users need to know how what inputs to put into a prompt, and tune the results until they get to a result close to what they are envisioning. Only then will they become engaged users.*

AI interactivity is a motion of sense and respond. We can’t see under the hoods of what is happening beneath the interface. The only way we can get better results from AI is by first getting bad results from AI.

This is what I mean when I ask, [what if we reframed AI as discovery?](https://www.linkedin.com/posts/emmiecampbell_design-productmanagement-artificialintelligence-activity-7166229805212819456-3wUi?utm_source=share&utm_medium=member_desktop)

That’s what users of digital products are encountering. BUT – they are used to predictable results, not working with the computer to product a result together.

We suffer from [automation bias](https://www.perplexity.ai/search/Tell-me-about-pI3XIV8NQc22XuxJn4zAtg). We trust the machine. So if the machine gives us a bad results, or incorrect results, it feels dissonant. It’s uncomfortable. And in many cases we don’t know how to respond. We don’t know that we CAN respond.

It’s the responsibility of the AI to guide users forward in their journey. AiUX must move people forward. It’s not sufficient to provide stuff and then fail to provide a way to improve that stuff.

There are emerging patterns that work this way.

For example, [nudges](https://www.shapeof.ai/patterns/nudges) early in the journey can help users construct stronger prompts up front. [Templates](https://www.shapeof.ai/patterns/template) that show users how [parameters](https://www.shapeof.ai/patterns/filters-parameters) can help them tune their output teach them how the technology works while creating a good experience from the start.

## The Role of Designers in an AI World

Designers working with AI are not just creators of visually appealing, usable interfaces. Designers need to be architects of experiences, curators of human-machine interaction, and, importantly, mediators between complex algorithms and users.

Our role is to make AI accessible, usable, and beneficial by translating its capabilities into applications that address real human needs and preferences.

In addition to adopting new AiUX interaction patterns, we need to revisit our patterns of work as well.

- **Treat users like co-creators**: Designers can no longer be on the receiving end of customer insights processed by someone else. We must make time in our process to get close to the people using the AI. Their unspoken preferences will manifest in their results. We need to be prepared for them.
- **Learn like a liberal arts major**: As interactions with technology get closer to feeling human-like, liberal arts disciplines can provide critical perspectives on how humans interact with technology. Psychology, Anthropology, and even philosophy teach us how human behavior adapts to unseen inputs and forces, similar to how AI will adapt to its inputs, we will collectively adapt to the presence of AI.
- **Think like a scientist**: How can we get out of the cycle of production. We need to be more comfortable viewing failure as a learning opportunity and not as a setback. Remember, improving our interactions with AI starts with bad interactions. This is the same as with humans. We learn emotional intelligence, social skills, etc through uncomfortable experiences as much as with comfortable ones. Designing with AI will be emergent and chaotic. We need to adapt our practices to accept that.
- **Prototypes as a super power**: I’m stuck on this one. What do prototypes look like in AI world. If nothing else, designers must get comfortable with prompt engineering. We need to understand how the computer is talking to itself when its returning results to people. If you haven’t read [John Maeda](https://www.linkedin.com/in/johnmaeda/)‘s book [How to Speak Machine](https://www.amazon.com/How-Speak-Machine-Computational-Thinking/dp/039956442X), stop what you are doing and go buy it. Rapid prototyping will be critical to help us fail and learn fast. We must learn the rules of computation to expand this skill.

## My vision

**The Shape of AI exists to help make the technology and impact of artificial intelligence more understandable, so that collectively we can influence a future where technology enhances our life, instead of causing harm.**

This is a work in progress. Every pattern I describe in this project represents a summary of current trends in usage, and best practices that are emerging – but from MY perspective.

I need yours!

I have tunnel vision. I’ve been staring at this for too long. If you see something you disagree with, or something I’m missing, I want to hear about it.

This list will continue to expand to include additional prompt types, UI patterns, settings options, and more. Additional pattern categories will include heuristics, evaluation and prototyping, and business use cases.

I’ve set up a [Slack](https://join.slack.com/t/theshapeofai/shared_invite/zt-2dydjb12p-Q3w38TSwJtJ98VPFt0NRQQ) for us, the Shape Shifters, the people changing ourselves so we can change the world around us.

Join, and share what you’re seeing.

This is messy, and emergent. It has to be. On the other side of this work is a new understanding of interaction design. A new definition of what it means to be an interaction designer. A new structure for design and discovery. A new operation plan for making it work.

## Building our shared future, together (a parting thought)

I spent some time over the weekend revisiting Christopher Alexander’s work. His “A Pattern Language” has become a critical reference for thinking about architecture in digital spaces.

But it’s another book in that series, “The Timeless Way of Building” that I think is the more critical reference in this moment. This is how he describes that book, and its importance to contextualizing the patterns in “A Pattern Language”:

![Quote: It is shown there, that towns and buildings will not be able to become alive, unless they are made by all the people in society, and unless these people share a common pattern language, within which to make these buildings, and unless this common pattern language is alive itself](https://emilycampbell.co/images/blog/a-new-pattern-language-for-a-new-paradigm-shift-img-2.webp)

Language shapes us, and we shape it. This is true even for our digital languages.

The way we talk about these patterns, the purposes we ascribe to them, the connectivity between them, defines the experiences where they are used.

We must develop a shared language if we are to build this future paradigm together.

Thanks for reading. Shoot me a message with your thoughts, [visit the site](https://www.shapeof.ai/), and [join the Slack](https://join.slack.com/t/theshapeofai/shared_invite/zt-2dydjb12p-Q3w38TSwJtJ98VPFt0NRQQ). I look forward to hearing from you!

❋

Emily

---
*Originally published at [https://emilycampbell.co/writing/a-new-pattern-language-for-a-new-paradigm-shift](https://emilycampbell.co/writing/a-new-pattern-language-for-a-new-paradigm-shift) on 2024-02-29.*


# The Shape of AI

*We are only just beginning to see the form that this new technology is taking in our products and our experiences. While we develop the patterns to design for it and with it intelligently, we must also be aware of how it is shaping us in return.*

### A paradigm shift

If you are reading this, chances are you think about software, or digital experiences, or content a lot. Maybe you’re a designer, or an engineer, or a content writer. Maybe you work for a company that designs or builds. Maybe you’re a founder.

No matter who you are, you’re probably noticing that things are changing, fast.

And yet, not as fast as you might expect.

We are seeing new features using AI pop up, but they mostly feel the same right now. Chatbots that you can direct, summaries of long articles or transcripts, pictures that look cool but maybe feel a little off…

Even the companies that are popping up in this space seems like they are solving similar problems, or solving different problems similarly.

Either AI is totally overhyped or we are standing of the edge of what is sure to be a very big cliff, with no idea of how deep it goes.

I think there’s no way it’s the former.

But I also don’t know what’s over the edge. I don’t know how much this change is going to effect our daily lives, our jobs, our relationship to technology, and the role technology plays in the products and experiences we interact with constantly.

I’m paying attention though.

## Slow at first and then all at once

Even if technology feels overwhelming at times, up until now, and for the most part, we have felt in control. We can blame bots, and big tech, and bad actors, and even “the algorithm” for the things we don’t like, but until recently I rarely talked to someone who felt like they truly didn’t know what their computer was doing.

That started to shift a bit as voice devices like Alexa appeared, raising fears of ever-listening machines. Waves of misinformation in recent political cycles raised the alarm.

But then in the last 6 months or so it seems like all hell has broken loose.

Is that post written by a human or ChatGPT? Is that photo of your house real? Did that politician really say that, or did that celebrity really endorse that product? Will I be laid off tomorrow when a computer takes my job?

Can we control *any* of this?

Now I’m hearing questions like this daily. I don’t have the answers to them, but we can find clues to those answers in the design of the products and experiences we interact with.

Take this platform where I published my newsletters. Hidden deep in the settings, half a page down, turned off by default, is this option:

![](https://emilycampbell.co/images/blog/the-shape-of-ai-img-2.webp)
*A screenshot of a setting from Substack’s writer dashboard reading: "Block AI training — This setting indicates to AI tools like ChatGPT and Google Bard that their models should not be trained on your published content. This will only apply to AI tools which respect this setting, and blocking training may limit your publication’s discoverability in tools and search engines that return AI-generated results."*

Think about the implication of that. Substack is incentivized to take my words, feed them (probably for a kickback) into the model of large tech companies that will sell it back to other consumers, reproduced as thoughts that those people can post on their newsletter.

Without crediting me, without verifying if what I am writing is true.

Alright, cry me a river, so someone could steal your newsletter. Sounds innocuous, if not innocent…

But what if that was a photo of you instead, hidden deep in Instagram’s settings. What if it was your child’s voice, borrowed from voicemails they left you. What if it was your life’s work?

Small design choices will have massive impacts on our sense of personal ownership, our trust, and our relationships to each other – relationships increasingly shaped by the content and experiences we share online.

### Form and function

This is what I’m curious about, and what I’m going to share here.

I’m a designer. My profession is to understand how people interact with each other, products, organizations, and institutions, and to nudge behavior through content and software to make those interactions better.

Hopefully, better for the person using them, not just for the company, institution, what have you.

As computation (AI) changes these relationships, these interfaces and interactions will change too. Our design needs to change with it. What we design, how we design, where we design, who we design with.

And who designs. Or, without getting to the “everyone is a designer” debate, let me restate this: Who takes ownership of the end experience. Who influences the end experience in an intentional way. Prompt engineers, model builders, business operators, and more will all have an impact on how this technology affects people, and what incentives or loops are built into the products they interact with.

You may be familiar with the phrase “form follows function.” In non jargon, this means the way something works should be based on what I want to do with it.

Right now form is driving function. We are in the early stages of this technology impacting our world. We don’t know enough about it to push yet back.

But we’re starting…

### Future topics

I chose the title of this blog intentionally. The I’m going to try and balance the content between patterns in the wild, patterns of work and experience, and the occasional deep theoretical posts (which I’ll try to minimize)

Some of the topics I plan to write on include

- Emerging visual and interaction patterns in AI products
- How it’s shaping how we work
- The behavioral and societal impact of this technology entering our lives
- How we can lead teams and organizations through these changes
- How we need to shape ourselves with new skills and mindsets

Send me a note if you have other ideas, and please subscribe to follow the newsletter. Better yet, share it. The bigger this community is, the faster we can move together.

---
*Originally published at [https://emilycampbell.co/writing/the-shape-of-ai](https://emilycampbell.co/writing/the-shape-of-ai) on 2024-02-22.*


# Trust is the currency of change

Trust is the currency of change.

I’ve used this simple concept for years to help teams and individuals think critically about how they are positioning themselves to have a bigger impact.

The only thing constant is change, especially now.

## First-order thinking

As **individuals** — how do we make sure that as we change (learn, grow), we are building trust to enable us to reach bigger heights: promotions, raises, new projects, moving into management, etc. How are we feeding the cycle?

As **companies** — as our products and experiences evolve (read: introducing AI! Pivot or expand.) we continue to build and maintain the trust of our customers so they are willing to come on the journey with us, and take bets on our future in the form of expanded contracts, recommendations to others, and engaging in the process.

As **leaders** — how do we bring people along so they feel like change is exciting and can benefit them personally, instead of feeling like change is chaotic and hoisted upon them?

> [!note]
> Invest time in building trust, so that when you need to spend it, it’s there.

## Second-order thinking

This is sufficient for maintaining your growth path. What do you do when you want to jump up a level? Don’t just focus on what others are doing who have the job/customer base/influence you want. Look at how they are building trust, and invest yourself in those activities.

- **Individuals** — Take initiative to perform the tasks that would be asked of you in the new role. Build a network. Track your impact.
- **Companies** — Make sure your end-to-end customer experience is solid across their lifecycle. Invest in community. Tell their story.
- **Leaders** — Understand the problems of the people around you, above and below, and make sure you’re helping to solve for them. Ask for help. Actively listen.

Trust is what moves us forward. It’s what we spend to get new shots on goal.

Invest in it first. Change will follow.

---
*Originally published at [https://emilycampbell.co/writing/trust-is-the-currency-of-change](https://emilycampbell.co/writing/trust-is-the-currency-of-change) on 2024-02-15.*


# Using design to manage risk

The conversation was swirling. The executive was trying to make a decision, but leader A and leader B were talking past each other. At first, I held back to listen…

Then I saw the underlying problem.

We had to make a decision about whether to sunset a legacy feature, but the executive didn’t have a full understanding of what the current experience of that feature looked like, and what would change. The other leaders were trying to *convince* him of their perspectives, but they were doing a poor job of explaining what the tradeoffs were.

He couldn’t assess the risk, so he couldn’t make a call.

I opened Notion and started a new table. About 15 minutes later I had a rough user journey mapped out: 6 columns, three rows: Current state, Future state A, Future state B.

It wasn’t colorful, it wasn’t fancy. It got the job done.

I grabbed 15 minutes of his time and sent the table ahead with a short explanation.

“This is exactly what I needed!” he exclaimed. We had a decision by the end of the day

As designers we forget that our tools work for us. If I had taken 2 days to make it pretty, or complicated, the opportunity to provide impact would have passed me by. What I created was discardable. No one ever looked at it again. But those 15 minutes paid dividends in time saved.

People struggle to solve a problem they can’t visualize. As designers, this is one of our super powers. Help people see the maze. Give them a map so they can find their way out.

The value of design is sometimes as simple as helping someone see what they can’t on their own.

---
*Originally published at [https://emilycampbell.co/writing/using-design-to-manage-risk](https://emilycampbell.co/writing/using-design-to-manage-risk) on 2024-02-11.*


# Questions to ask to get more out of discovery

If designers aren’t participating in discovery, they are missing out on understanding what is really driving users. Here are some questions I coach designers to think about while conducting discovery:

👉 Hint: You will never hear me ask “do you like this” 👈

- What is the outcome they are trying to achieve?
- What does it mean for them to have achieved that outcome?
- Are there minor outcomes they seek to achieve along the way (base camp 1, base camp 2, summit)
- How do they currently try to achieve that outcome?
- How well do those alternatives work for them?
- What is their risk/tolerance to switch from those alternatives?
- What have they tried in the past that has failed?
- What have they wanted to try but haven’t/couldn’t?
- What happens if they don’t achieve it?
- What hacks have they put into place to achieve it?
- What hacks have they put into place to account for not achieving it?
- How much pain would they be willing to take on to break those hacks?
- How much pain would they be willing to take on to adopt new behavior?
- Who else needs to be involved for them to change that behavior?
- What processes need to change or be invented to change that behavior?
- Do they have decision making power around those processes?
- If there was one thing that could tell us we were on the path to helping them achieve this outcome, what would that signal be?
- What is the ultimate signal?
- Is this even what they are really wanting to achieve? Are we thinking too big? Too small?
- How critical is this to them? (For their sense of purpose? Accountability? Financial health?)
- When they look around, what is an example of someone that has achieved this? How did they do it?
- What existing tools remind them of the solution?
- What tools seem like the wrong solution?
- What are the anti-patterns?
- What would I expect to hear them say if we found the right solution?
- How would they pitch it to others?
- What would it take for them to invest time/social clout/risk/money into this solution?
- What would it take for them to invest it? Why wouldn’t they?
- Why do they care?

---
*Originally published at [https://emilycampbell.co/writing/questions-to-ask-to-get-more-out-of-discovery](https://emilycampbell.co/writing/questions-to-ask-to-get-more-out-of-discovery) on 2024-02-04.*


# Navigating the unknown

For all of the debate recently about the value of design, I rarely see designers talk about the merits of speculative design.

While we may measure the value of traditional digital product design against how design helps us meet our existing targets, speculative design is a strategic tool that can unlock new business opportunities and drive innovation in product development. Far from being a mere creative exercise, it challenges designers to think beyond the current market needs and user behaviors, enabling them to anticipate and shape future trends.

I've experienced firsthand how this approach can help unlock millions of dollars in new revenue and strategic partnership, delivered through design. Unfortunately, due to lack of exposure or understanding, this approach to leverage design is often misunderstood or underutilized by designers.

In an uncertain world, speculative design is a game changer, and an advantage for designers to adopt to stay ahead of the curve. But what exactly is speculative design, and how can it be effectively integrated into SaaS product development?

## What is Speculative Design?

Speculative design is a forward-thinking approach that goes beyond solving existing problems. It's about exploring 'what could be' rather than 'what is.'

This approach encourages designers and developers to envision future scenarios, challenge current assumptions, and experiment with potential outcomes. Unlike traditional design methodologies that focus on immediate solutions, speculative design is about questioning, probing, and imagining.

When applied in a business context, it helps us think beyond our current strategy and solutions and bring others with us. An example might be the prototype of a concept, not yet built or validated, that shows a potential future partner what a possible integration might look an feel like. It helps tell a story, and excites others around a vision that they might not have been able to picture.

In effect, speculative design is a means of 'pinging' the market, like radar helping us sense opportunities we cannot see yet.

## How to Use Speculative Design in Product Design

In the realm of digital product design, where user needs and technological capabilities are constantly evolving, speculative design offers a way to anticipate and prepare for future trends. It's not just about creating a product that meets today's market demands but about envisioning how these demands might change and how technology might evolve.

- **Start with Scenario Building:** Begin by imagining various future scenarios for your product. Consider factors like emerging technologies, potential changes in user behavior, and broader societal trends. What challenges and opportunities might these scenarios present?
- **Encourage Open-Ended Exploration:** Speculative design thrives on creativity and open-ended questions. Encourage your team to think beyond current constraints and explore a range of possibilities, no matter how far-fetched they may seem.
- **Focus on User Interaction:** Consider how future users might interact with your product. How might their needs and behaviors change? Use speculative design to explore new forms of user interaction and engagement.
- **Prototype and Test:** Develop prototypes based on your speculative designs. These don't have to be fully functional but should be enough to convey the concept and test its feasibility or desirability.
- **Gather Feedback and Iterate:** Use feedback from user testing to refine your speculative designs. This iterative process can reveal insights into current user needs and potential future trends.

## Examples of Speculative Design in Practice

### Finding the Perfect Fit: Speculative Design in Market Positioning

Speculative design acts like a puzzle piece in the quest for product-market fit, helping companies identify their place in a complex market landscape.

*At Degreed, I used speculative design to explore how the company could assist a major client in rethinking the connection between jobs and skills, aiding employees in navigating opaque career paths. This exploration led to unlocking millions of dollars in future revenue by addressing a critical market need.*

### Fostering Innovation: Speculative Design as a Catalyst for New Ideas

Speculative design serves as a navigational tool in uncharted market territories, especially when traditional constraints and incentives are not well-defined, allowing companies to explore and understand emerging consumer needs and market trends that are not immediately apparent.

*At Vendr, I've used speculative design to demonstrate how AI might help people discover insights about their SaaS contracts and quickly un-tap ways to save money. This helped us identify new ways of thinking about our data and delivering customer value that was not obvious without this exploration.*

### Attracting Early Interest: Speculative Design as a Market Teaser

This approach is instrumental in generating early customer interest and attracting business investment. By offering a glimpse into potential future products or services, companies can gauge market reactions without fully committing resources.

*At BookClub, I built a robust prototype for a B2B-oriented offering that people could play with in a real world setting. This helped us better understand the market opportunity and led to customers investing in our idea as early adopters.*

### Risk Mitigation: Speculative Design in Anticipating Challenges

By anticipating future challenges and trends, speculative design can help companies mitigate risks and adapt to potential market shifts.

*I often combine speculative design explorations with pre-mortems, working backwards from a concept to identify the biggest risks we need to learn more about or validate a solution against before over-investing in a potential dead end.*

## Conclusion

Speculative product design isn't just about creating; it's about learning, adapting, and evolving with the market. It's a journey of discovery, where each step is an opportunity to learn more about the market and the consumers within it. As designers and innovators, embracing this approach could be our key to unlocking potentials we never knew existed.

---
*Originally published at [https://emilycampbell.co/writing/navigating-the-unknown-how-speculative-design-unveils-business-value](https://emilycampbell.co/writing/navigating-the-unknown-how-speculative-design-unveils-business-value) on 2024-01-04.*


# The Discounting of Design

'Tis the season for resolutions. Personally, my resolutions this year include writing more, putting away my devices when I'm around my kids, and strengthening my core.

As I reflected over the last few weeks about my intentions and goals for the upcoming year, I began to think about value, and how we calculate it. After all, a resolution is just a decision to increase the value of something I might get later over the value of something I can enjoy today, right?

I choose to value stronger core muscles (something it will take a bit of time to realize) over lazing on the couch and watching a movie instead of working out (something I can enjoy today).

In economics, this relates to a concept called *discounting*. Generally speaking, we tend to value things we can have today over things we can have tomorrow. In other words, we *discount* the value of things in the future.

We do this for all sorts of reasons: the future is less certain, the value of future things may change, etc. A high discount rate means something loses value faster the further its result or impact is out in the future, whereas a low discount rate means I retain its value to me compared to things I can get today even if I won't get that result for a while.

I place a lower discount rate on gold, for example, because I assume it will retain its worth even if I hold onto it for a long time. I'm less likely to spend gold in my savings in exchange for something in the short term, like a new coat. However, if that savings is kept in dollars, I might be more likely to dig into it to buy that fancy treat for myself for the holidays, because I have less trust that investing in it will have a high return.

*A resolution is a mental game to lower the discount rate of some future value.*

So what do discounts have to do with design?
![](CleanShot%202026-06-03%20at%2015.43.46@2x.png)

There are three primary points that have been swirling around my head over the holiday break.

## 1. Valuing outputs inflates the discount rate of design

Like our peers in engineering, design has been valued for its short term gain: its outputs of production.

The more a designer could produce in the same period of time, the higher their value. If a designer making $60,000 annually could produce $80,000 worth of work, you would say that the Annual Return on Investment (ROI) for that designer is 133%.

Most people have a cap on the amount of work they can do in a day, and that cap is not likely to increase dramatically as they gain experience. So the ROI of design in these environments is relatively fixed and linear.

These types of businesses hold little incentive for designers to engage in work outside of producing the outputs of design. For every hour that a designer is doing research, that's one hour that they aren't producing, decreasing their return.

And because these businesses don't have proof that some future work based on design research will be significantly better, they have no reason to ascribe a higher ROI to design practices.

Factor in the fact that I discount the value of that future work against today's work and design research starts to look like a net loss.

**The discount rate means that when design is valued primarily for its outputs, there is no business incentive to invest in design practices that support future work over work today.**

> [!note]
> Because of the inflated discount rate of design in an output-driven environment, the more design focuses on production, the less incentive there is for businesses to invest in the practices that lead to better design outcomes.

The problem with this incentive system is it's based on a fallacy.

When design is applied as a practice, and not just a production role, it actually creates more value for the business. We see this in more mature organizations: design that is holistically integrated and applied across business processes (e.g. not just toward producing artifacts for development), the more value the business generates.

This is backed up by [multiple](https://www.invisionapp.com/design-better/design-maturity-model/) [studies](https://www.mckinsey.com/business-functions/mckinsey-design/our-insights/the-business-value-of-design). Design practices like design thinking, design research, and design strategy, have a positive impact on the bottom line.

But in order for businesses to recoup this value, they must transition the way they deploy and fund their design resources. This may lead to a temporary dip in the perception of the value of design as design teams re-orient their production work in order to focus on higher-value practices.

A great example of this is when leadership underfunds or puts off the creation of a design system, which they justify because it would require at least one designer to shift their focus from production work. However, we know that this short term decrease in production in order to give the team time to organize an effective system can actually increase velocity and value in the future, well beyond what is possible without a system in place.

Leaders continuously put off investing in work that will lead to faster and more consistent value after some period of work, because they presume that interim work is wasteful.

**Therefore, there's a contradiction to design value: the more teams focus on creating short term design outputs, the less likely they are to realize the full value of design.**

## 2. The ROI of design practices is a multiplier, not a fixed return

Adding to the logic above, businesses caught in this fallacy act as though the ROI of design practices is the same as the ROI of design production. That simply is not the case.

When designers are focused solely on production, we see that the return on their work is fixed. However, when designers balance their focus on production with expanding the human-centered, creative mindset across an organization, they can positively impact far more people and processes than they would if those hours were only spent producing.

This network effect suggests that the ROI of design practices is more of a multiplier than an additive number.

I don't claim to know what that multiplier is (…yet, but I want to find out). But if this thinking is correct, then we're currently talking about the value of design all wrong.

**The value of design practices is the ability to increase the efficacy, decrease the risk, and scale the impact of work across the business by a factor of x.** X increases as the maturity of those practices expands (and not necessarily just the number of people doing them increases).

Therefore, the business value of design is directly related to its maturity and its multiplier effect, and will be seen not by looking at the outputs of design itself, but by assessing the outputs of the functions and outcomes that design influences.

***But the business must resolve itself to get there.***

---
*Originally published at [https://emilycampbell.co/writing/the-discounting-of-design](https://emilycampbell.co/writing/the-discounting-of-design) on 2020-01-04.*


# What product designers can learn from Motown

I'd always thought that Motown's success was a combination of its talented stars and fortunate timing, but I [only recently learned](https://www.sho.com/titles/3470002/hitsville-the-making-of-motown) that much of its success was due to a mastery of the power of the team–loosely coupled, highly aligned. Everyone at Motown was empowered with a singular goal that they all collected around: to create hit records.

And the mindset and methods responsible for Motown's huge success in its heyday are just as relevant today.

These methods allowed Motown to build a powerful culture that turned an $800 investment in 1960 into $20 million annual revenue within six years and produced some of the biggest hits and superstars of the 20th century.

But teams that want to generate their own success don't need a Diana Ross or Smokey Robinson to make it work. Motown capitalized on its vision by following a set of first principles that will sound quite familiar:

- Iterate rapidly through divergence and convergence
- Accept failure as a necessary step to learning
- Critique artifacts, not people
- Scale through strong, service-based leadership

## Diverge, Converge, Repeat

Before starting Motown, founder Berry Gordy worked in a Ford-Lincoln plant in Detroit. It was there that he learned about production lines and quality control, though his company ended up resembling more of a modern design studio than a Model T factory.

Just like a design studio, constraints often produce the best results. Motown consistently created market impact by balancing divergent and convergent thinking. Divergence allowed them to produce a huge stock of ideas, and convergence helped them narrow those ideas down objectively to the few most likely to succeed.

**A typical weekly process looked like this ([as told by former engineer Bob Ohlsson](https://web.archive.org/web/20160508203442/http://bobolhsson.com/bob-says/on-motown/)):**

- Each writing/production team was required to generate five new songs a week
- During the Monday-morning production meeting, the two strongest submissions from each team would be identified, and the teams would focus their week on getting to a basic recording of each
- Each team would generally take only one song to finished quality
- Recorded songs went through several mixes–some received over 100, though the average was 15
- At the end of the week, the quality control department listened to each final recording and determined whether it would be a hit. No song could be released until the department unanimously agreed it was ready

This feels familiar: apply limited gates early on to encourage open ideation, and gradually narrow through cross-functional critique and prioritization. This has the benefit of removing ownership bias while moving risk upstream.

> The idea was to have more flops within the company and fewer on the streets
>
> — Bob Ohlsson

The company invested time early in the process to produce lots of ideas when risk was low. By final release, over 20 songs and over 10 recordings had been discarded (each week!). Though this process was expensive, it was less costly than producing riskier albums that were more likely to flop after distribution.

**Consider how you could save time and money by adopting a similar process of iteration and critique:**

- Make creation a habit: Hold regular design studios and invite non-designers to participate in order to build the skill of rapid idea generation
- Work small, think big: Batch problems and ideation into small iterations so ideas don't get stale
- Build critique as a muscle: Hold regular critique sessions and force yourselves to take only the strongest ideas forward
- Gate decisions, not ideation: Empower individuals to generate and explore ideas independently, and set clear standards and principles as a unit that every idea must meet to move forward

## Adopt a Culture of Learning

The story behind the single "Shop Around," by the Miracles stands out as an example of their culture at play. Two weeks after the song was originally released, Gordy called up songwriter and lead singer Smokey Robinson at 3am to say he thought the song was too bluesy; it needed to be more upbeat for wider commercial appeal.

Despite the sunk cost (the vinyl had already been pressed and released), Gordy assembled the band in the middle of the night at the recording studio. The re-released track shot to the top of the charts and became the first Motown single to sell a million records. Gordy had to be willing to accept risk or failure. he assumed he didn't always have the right answer, and that the first answer wasn't always right.

The flipside of this occurred over a decade later. After Marvin Gaye recorded *"What's Going On"*, Gordy apparently [called it the worst song he had ever heard](http://content.time.com/time/arts/article/0,8599,1870975,00.html) and shelved it.

Gaye trusted himself and his music. After refusing to produce additional songs until his was released, an executive at the company helped release it without Gordy's permission. It became the fastest-selling single the label had ever produced.

Gordy quickly acknowledged that he was wrong, and gave Gaye creative freedom to produce the entire album. His concession to let the artists take more control over their work led to the success of Stevie Wonder, and the series of soul and funk-based artists that carried the label into the next decade.

**A culture of learning balances risk with experimentation**

Motown maintained short iteration cycles, balanced data with instinct, and kept a culture that allowed individuals to challenge leadership. This combination resulted in an agile and adaptive process that persisted for decades.

## No Room for Ego

Berry Gordy discovered quality control while working on an assembly line, and applied the concept ruthlessly at Motown. His *Brain Trust* meetings had one objective: identify which songs were most likely to be successful.

> The goal was to create a hit record by suggesting how the best of the week's work could be improved
>
> — Berry Gordy

By creating a specific and common purpose for everyone involved, critique was expected, and never personal. Gordy actively created "[an atmosphere of safety of ideas and thoughts](http://www.oprah.com/own-master-class/Berry-Gordy-Describes-Motowns-Private-Internal-Meetings-Video#ixzz2vsXjGmo9)" where anyone could voice their opinion without fear of retribution.

Critically, he was not immune from the process. Despite his role as founder and CEO, Gordy continued to write and produce songs, but his position offered no guarantee that his songs would be selected to be released. This added to the culture of trust and candor that allowed for a democracy of ideas to flourish.

Gordy didn't surround himself with yes men or music producers in these meetings. Instead, he filled the quality control department with a set of people who could speak to the entire cycle, from production to distribution to market trends.

> A non-creative person's vote counted just as much as a creative person's. I took the democratic approach because although I was in charge at Motown, I made logic the boss: no egos or politics allowed. Not even mine. And I did it because of truth. "The truth is a hit," was what we used to say in our Quality Control meetings at Motown.
>
> — Berry Gordy, *To Be Loved* (1994)

**Think about how you might adopt a similar Brain Trust in your organization:**

- Open it up: Make inclusion a priority to avoid group-think and ownership bias.
- Don't make it personal: Everyone should be subject to critique and held to the same standards.
- Promote decisiveness: Everyone gets only one vote, and the question they are voting on should be clear. In Motown's case, the question wasn't "do I like it" but "will it be successful".
- Use real data: How can you keep your perception from skewing decisions? Berry avoided this by playing songs through a tinny car stereo to reflect a realistic listening environment, so they could hear what the customer would hear.
- Demand excellence: Hold everyone accountable to the group's goals. The Brain Trust collectively was accountable for the success of the hits that were released to the market, and they ensured everyone from artist to salesperson was supported and responsible.

## Leadership is Culture

Much of Motown's ability to learn from mistakes and successes alike owed to the strong culture of accountability and shared ownership. Gordy demonstrated this as a leader, creating a space for others to feel empowered to follow. "If they don't get a hit, it's our fault, not theirs," he said of the artists.

There was no question who was in charge. Gordy gave himself power of veto power at all quality control meetings, and remained involved in the day-to-day operations of the label, even as it scaled.

However, he made two crucial decisions on how to lead that cultivated the creative atmosphere around him.

First, he hired people who were better than him, better songwriters, better producers, and he refused to be intimidated by them. This challenged everyone at the company, including himself, to adopt a learning mindset that led to personal growth.

Second, he allowed for complete creative freedom in the discovery and production of ideas, while maintaining strict, collective principles that everyone adhered to for decision making. This allowed him to distribute risk and accountability to create a sense of co-ownership that drove the entire company toward excellence.

> Hitsville had an atmosphere that allowed people to experiment creatively and gave them the courage not to be afraid to make mistakes. In fact, I sometimes encouraged mistakes. Everything starts as an idea, and as far as I was concerned, there were no stupid ones. "Stupid" ideas are what created the light bulb, airplanes and the like.
>
> — Berry Gordy, *To Be Loved* (1994)

Many companies struggle to find a balance between power and autonomy, but Motown proved the model. While Gordy maintained control over the direction of the company, he used his power to empower others, and then granted them the freedom to do something with it.

Making hits was the only thing that mattered. Because the Motown north star was so precise, it left no room for ego. Decisions could be made on principle, stars had the same accountability as secretaries, and even the man at the top wasn't immune to criticism.

That's the secret to Motown's success.

**Your leadership will reflect in your organization's culture. Expose your leadership to the same critique as you would other parts of the culture to test its strength:**

- Hire to your weaknesses: Ask the team to assess your relative strengths and weaknesses through a 360-degree evaluation.
- Make decisions clear and democratic: Ensure your organization has a clear vision and principles to help you make decisions objectively, and ensure everyone, including leadership, is accountable to them.
- Expose your blind spots: Ask others to write out the [unwritten rules](https://hbr.org/2019/10/why-you-should-write-down-your-companys-unwritten-rules) of your organization's culture to expose your invisible culture.
- Conduct research: Hold regular interviews to assess how empowered your team feels to act on and own collective goals.

## It's in the Grooves

Berry Gordy knew that culture existed [in the grooves](https://www.rollingstone.com/music/music-features/the-motown-story-how-berry-gordy-jr-created-the-legendary-label-178066/): in the often invisible patterns that exist between people. For pop music, these were simple themes of love, joy, and heartbreak. In his company, these were the defined patterns that everyone would follow so they could focus their energy on creativity and precision.

Whether you adopt all of his methods or some, Gordy proved that an organization could be successful through empowered teams, rapid iteration cycles of creation and critique, and a shared purpose.

If you want to learn more about Motown, check out the following resources:

- [Hitsville, the Making of Motown (Showtime)](https://www.sho.com/titles/3470002/hitsville-the-making-of-motown)
- [Deep Soul, the Rising up of Motown](https://www.youtube.com/watch?v=wkd1c4T5HiE)
- [To Be Loved, Berry Gordy's memoir](https://www.youtube.com/watch?v=wkd1c4T5HiE)

---
*Originally published at [https://emilycampbell.co/writing/how-motown-built-their-success-on-a-culture-of-creativity-and-collaboration](https://emilycampbell.co/writing/how-motown-built-their-success-on-a-culture-of-creativity-and-collaboration) on 2019-12-31.*


# Inclusion, joy, and superpowers

There’s currently a[ video circulating](https://www.si.com/more-sports/2019/02/08/miles-taylor-cerebral-palsy-deadlift-weight-lifting-video) of a man with cerebral palsy dead lifting 200 pounds, followed by a raw outburst of joyful celebration from his friends and supporters.

Lifting weights like that is an incredible feat for any human, but the moment is made more special by the fact that the man only weighs 99 pounds himself.

Since most of us don’t live with cerebral palsy, and many of us aren’t exposed to anyone who does, this video offers a rare moment to challenge our assumptions about the limitations of that condition and connect to the people who experience it.

As I spend more time focusing on inclusive design, I’ve become acutely tuned in to noticing the way we celebrate small acts of inclusion: The gym teacher pausing to redo the hair of the young student in his class, Susan Boyle sending shocks through the audience, the student athletes who help an injured peer to the finish line.

They create joy for the individual, and for those around them. Creating a more inclusive environment generates community, compassion, belonging.

This morning I read a [phenomenal article from the New York Times Magazine](https://www.nytimes.com/2019/02/05/magazine/letter-of-recommendation-color-blind-pal.html) on colorblindness. The author describes all the times she was misunderstood due to her visual difference, and expresses pure exhilaration over an app she discovered for her phone that allows her to sense color.

She exudes that when she pointed the app at a red sticker, “The color felt so vibrantly *red*.”

“Felt” Red?

I don’t think I feel color, but I also think that’s my weakness and not a strength.

Her super power is tuning into the world in a way that others can’t, helped by technology in this case, because of a physical difference that others have belittled her for.

But despite this incredible capability that the app loans her, the feature she is most excited about is the one that allows others to sense the world as she does.

> The feature’s most striking effect is the ability to inspire a strange kind of empathy. Last summer, I took a trip to the Southwest, a part of the country known for the bold reds of its adobe buildings, its mesas, its sunsets, its sands and cliffs, its everything. To me, it all seemed pretty much beige…

> I switched to the setting that emulates my colorblindness and showed it to the man I was with. He looked at me with a mixture of perplexity and pity, as if to say: *This?* This is how you experience the world?

> From [“Letter of Recommendation: Color blind pal”](https://www.nytimes.com/2019/02/05/magazine/letter-of-recommendation-color-blind-pal.html) by Zoe Dubno in the New York Times Magazine

The joy of allowing others to experience your world is more impactful than the joy of experiencing the world as others.

As much as we all want to belong, every one of us has unique differences. Some are physical, some are intellectual, some are emotional; some are visible to others, most are not.

When we talk about designing for “experience”, I don’t think we reach far enough. We can create joy through beautiful illustrations, personalized content, direct connection, and so on.

But the real joy, the real opportunity for design, is the ability to connect others in a way that is not possible without what we create.

Technology has unlocked our ability to live as our authentic selves, instead of carving space in the realm of others.

My favorite TED talk is by Neil Harbisson, who was born with a rare form of color blindness that causes the visual world to be rendered in greyscale. He partnered with a team of scientists to create a device that would allow him to hear color.

What began as a disability has become his super power: The ability to sense the world in a way that no one else can. And he was assisted by technology, by design, and that’s our opportunity.

Through inclusive design, we have the privilege of helping people to discover their own super powers, to live authentically, and to fully experience the world in radical ways.

We have the ability to bring joy into the world, and what an opportunity that is.

---
*Originally published at [https://emilycampbell.co/writing/inclusion-joy-and-superpowers](https://emilycampbell.co/writing/inclusion-joy-and-superpowers) on 2019-02-09.*


# Design and motherhood

When I was growing up, I didn’t dream about becoming a designer. I enjoyed creating, exploring, communicating, but I always figured I would settle on a more conventional career. To be honest, I don’t remember having a strong conviction to be a mother either. Yet here I am.

I became a mom around the time I started my first full-time, in-house position. We moved to California 6 months after I started, and a week later we learned we were expecting. You can imagine this threw a wrench in my plans.

I had anticipated spending my weekends exploring the coast, mountains, and forests of my new state, to balance the grind of the week that was already burning hot. Suddenly, I found myself swapping hikes for doctor’s appointments. Instead of grinding, I bargained with my boss to let me work from home a few days a week.

This was a hard time in my life. I felt tired, undervalued, and anxious about my visibility in the company. Oddly enough, though, the experience of becoming a mother made me a better designer.

I share this because I’ve seen others express concern that becoming a parent will negatively affect their career. Certainly, there are examples of toxic companies that only want to hire 20-somethings with no children or worldly commitments (though why anyone would fund leadership with such short-sighted expectations is beyond me). Parenthood teaches you ways of thinking about, reacting to, and experiencing life that you can never understand until you’re in it.

## Humility

I’ve never been good at asking for help. In an odd contradiction, it’s my fear of showing weakness that often becomes my biggest weakness. Combined with an almost stoic perfectionism, I often find myself burnt out and overwhelmed.

Having children forced me to come to grips with this sort of self-defeating behavior.

When your work meeting is interrupted by a toddler home with the flu, or you don’t notice that milk stain on your shirt until you’ve been at the office for a few hours, or you get behind in your work because your little one had night terrors, you’re forced to change your perspective on life.

The best example of this is the most wonderful news clip ever aired, where an interview with an analyst on North Korea is interrupted by his daughter bouncing into the room. This pretty much sums up life with kids. You’re on their schedule, and you might as well embrace it.

I find myself taking things less seriously after kids. I’m better at asking for help, taking criticism, and rolling forward.

After all, we’re only human. We make bad assumptions, we mess up, we’re often wrong. I’ve gotten better at sharing work earlier and taking feedback. I’m less likely to take things personally. I shout out when I’m blocked or stuck and collaborate with others to move forward.

In parenting, a plan is often broken, but the act of planning is invaluable. Same with design. Hope for the best, plan for the worst, and just keep swimming.

## Empathy

The whole time I was pregnant, I was expecting a massive transformation after my son arrived. I’m not sure what I thought I would experience, but suffice it to say, it didn’t work out that way. One day I was me, the next day I was the same person accompanied by this little being I was wholly responsible for.

If you’ve ever wanted to clean up after your drunk friend while deflecting obnoxious questions in a room with deafening, terrible music for the rest of your life, I highly recommend having kids. All those funny [books](https://www.amazon.com/Go-F-Sleep-Adam-Mansbach/dp/145584165X), [Tumblrs](http://www.reasonsmysoniscrying.com/), and [memes](https://www.buzzfeed.com/awesomer/reasons-kids-are-pretty-much-just-tiny-drunk-adults) about parenting…yup, they’re about accurate.

At first I relied on instinct to get through. Just when you think you’re about to snap (and I mean, totally lose it), this wave of calm sweeps over you as your infant looks up at you, gasping between blood-curdling screams, and you fall in love all over again.

But suddenly this little screaming thing is talking to you, asking questions, throwing fits over the most trivial things. Suddenly, you’re raising this little person that you struggle to relate to or understand, and you’re responsible for them.

I’ve practiced meditation and yoga for about 20 years now, so I come in with an advantage. Still, I could never have anticipated the challenge and beautiful opportunity that being a mom brought to the table. My practice in mindfulness has improved immensely as a result.

I find myself reacting more slowly. When feeling frustrated or confused about my sons’ behavior, I pause. I recognize that I cannot truly put myself in their shoes, and I start to trace through why they may be feeling the way they do. Tears about a popsicle may relate to a lack of control. Potty training accidents are met with embarrassment, but also self-doubt. Expectations shape the way they interact with me and their peers, and I’ve learned to anticipate how they’ll respond to situations based on what came before.

This situational awareness lends itself to design. I’ve become better at exploring the different emotional and physical contexts within which someone may interact with my work. I’m more aware of bias, including my own confirmation bias, and to challenge my initial reactions and conclusions.

I’m more intentionally aware of the needs and unique experiences of others, and this has traced to other parts of my life in ways I never could have expected.

## Structure

Finally, despite all the chaos. being a mom has made me more organized and deliberate with my time and energy. I’ve learned I can’t be everything for all people at all times, and I’ve come to grips with this.

When people share concerns about becoming a working parent, I often hear them express a fear that people will judge a decline in their work ethic and productivity. While this may be a perception of non-parents and toxic superiors, the reality, in my experience, is completely different.

I used to be a workaholic. I still am I guess, but I force myself to structure my day with more intention. I have to carve out time for me, my husband, as well as work and children, while still accounting for uncertainty.

See the first point above—sometimes things just don’t work out according to plan. Still, I’ve become more focused and productive during my work hours, and better at stepping away and redirecting my attention during the rest of the day.

When I’m with my kids, I try to focus on being present with them. I’m still working on this, but my goal is to be better at experiencing them, fully listening, and investing in each fleeting moment.

I’m not sure if this makes me a better designer, but I think it has made me a better employee and a better person.

This world moves so fast. One day you’re holding your 8lb child in a blanket, and the next he’s losing teeth and telling you about his friends. It’s sad but beautiful.

Everything is temporary. Every win. Every setback. Every compliment and every criticism. We have to balance awareness in individual moments with an appreciation for the journey. In parenting, design, and life, this is a very empowering skill.

Parenthood is a crash course in life skills. It’s an incredible journey, and I’m grateful to be on it.

*Note: Some of my closest friends have chosen not to be parents, and some would like to be but are not there yet or struggle. I by no means feel that being a parent is the only way to gain these skills. **Happy Mothers Day to all moms, and love to everyone.*

---
*Originally published at [https://emilycampbell.co/writing/design-and-motherhood](https://emilycampbell.co/writing/design-and-motherhood) on 2018-05-14.*


# Designing for Humans

There is a growing trend throughout the web community to embrace an understanding of behavioral science, and to apply its tenets to our designs. This progress helps us walk the delicate balance between providing an emotional and pleasurable experience for our users and communicating content and information through clear, intuitive patterns.

When the web was first developed, it functioned as a large database, a means of transmitting information from one server to another. Its design was, inherently, mechanic, and placed little emphasis on experience or enjoyment. However, its usefulness as a computing tool was quickly surpassed by its potential to connect. The act of browsing the web grew from a personal, targeted experience (one person looking for specific information) to a multi-user and multi-use phenomena (countless people across the globe exploring a myriad of information and interactions).

Today, the web is a primary means of communication, information-gathering, and enjoyment. Its users have as many interests, limitations, and characteristics as they have faces. I cringe when I hear web design referred to as a facet of [“Human-Computer Interaction”](http://en.wikipedia.org/wiki/Human%E2%80%93computer_interaction), or HCI. Computers are mechanical and thus unable to elicit emotional responses to their users’ needs. If we inject a personal element to our designs, then we can provide an emotionally-driven interaction that is more than just a series of inputs and outputs.

When designing for humans, we recognize the innate differences that each person embodies while accounting for the absolute similarities that all humans share: a sensitivity to group dynamics, emotional stimulation, positive feedback, and familiarity.

## The Clique Mentality

Try as we may to distinguish ourselves as unique, the scientific consensus routinely points to our willingness as humans to adopt a “herding mentality,” wherein our decisions are weighted by what our peers are doing. Indeed, as Nir Eyal aptly points out in [UX Mag](http://uxmag.com/articles/designing-to-reward-our-tribal-sides), the need to feel social connectedness informs our values and drives much of how we spend our time. Scientists have found that [there is a distinguishable range in a social movement at which this instinct kicks in](http://www.sciencedaily.com/releases/2010/10/101013163402.htm), outside of which our decision to adopt a product (or opinion, or trend…) is left more to personal instinct than group persuasion. Malcolm Gladwell famously refers to this as [the tipping point](http://www.nytimes.com/books/first/g/gladwell-tipping.html).

Knowing that this social phenomena exists leads us to dissect how popular sites use it effectively, and gives us direction to apply the successful attributes to our own projects. Facebook and Twitter are obvious examples. Their users interact with the content based on a “my friends like this, so I should too” mentality. This social validation creates a sense of trust. Trust is also achieved when content is posted by a trend setter—a leader in an individual’s network. Either way, the individual endorsement adds value to the content, and that value increases engagement, bringing return to the site owner.

There is no algorithm to when something tips, so agile designers have to build products that provide adequate incentive for the trend setters to participate while keeping all users engaged. There are a number of strong examples of this across the web:

- [Dribbble](http://www.dribbble) lists shots by popularity, and a popular shot can propel a user’s status skyward. Members of the community have an incentive to post high-quality content, be active in the community in order to gain followers, and post often to increase their chances of reaching the popular page.
- [Yelp](http://www.yelp.com) relies on social validation as its core value. If someone you are connected to recommends a restaurant, that recommendation holds far more weight than a static review, and thus makes you more likely to try it out. Then, there is less social risk to posting about the restaurant after your visit since someone you know has already created a review, making you more likely to return and interact with the site again.
- Alternatively, social media buttons may act as a negative social indicator. As noted by Oliver Reichenstein of [Information Architects](http://informationarchitects.net/blog/sweep-the-sleaze/), a low “tweet”, “follow”, or “like” count can communicate that your content is not worthy of your reader’s trust or time. A high count may be seen as a personal advertisement, which can be just as much of a turnoff. [Medium’s](http://www.medium.com) approach to the “tweet if you like it” button walks a perfect balance between class and effectiveness.

> That button that says ‘2 retweets’ will be read as: ‘This is not so great, but please read it anyway? Please?’

> Oliver Reichenstein of [Information Architects](http://informationarchitects.net/blog/sweep-the-sleaze/)

## Designing with Empathy

I am going to put it out there that empathy is the most defining of human characteristics, at least in so far that it distinguishes us from our robotic counterparts. It is that sensitivity to what others feel that forms the bedrock of our political, cultural, and social institutions.

Humans react to the [emotional](http://uxdesign.smashingmagazine.com/2012/07/18/the-personality-layer/) cues of their surroundings. Just as our actions follow a herding instinct, our emotions feed off of each other as well (e.g. “misery loves company”). Developing a content strategy that considers the user’s mood and disposition creates a personal user experience. As noted by John Caldwell of [UX Mag](http://uxmag.com/articles/when-and-where-to-%E2%80%9Cwoohoo%E2%80%9D), a consistent tone of voice makes a brand’s character believable and trustworthy. Without it, customers might have a contradictory impression of the brand.

An excellent example of a company who embraces a consistent, empathetic tone across their brand is [Mailchimp](http://www.mailchimp.com). Their online document [Voice and Tone](http://voiceandtone.com/) provides excellent insight into how they combine friendly, useful micro-content with a strong brand identity.

![screenshot of Voice and Tone, by Mailchimp](https://emilycampbell.co/images/blog/designing-for-humans-img-2.webp)
*Mailchimp’s document [Voice and Tone](http://www.voiceandtone.com) provides insight into how the company successfully combines empathetic copy with a strong brand identity.*

We humans react to empathy in product design as well. Comprehensive user research and planning leads to products that are honest, usable, and meet our user’s needs and expectations. This applies to visual design as well as back-end design.

- A site for children will likely have a whimsical, friendly tone, while a news or financial site will showcase a more formal visual aesthetic.
- Sites that are optimized for mobile may serve lower-resolution images and utilize a simple, stripped-down interface in order to reduce strain on the user’s bandwidth.

However, empathy requires more of designers than simply making decisions based on the user’s demographics or other characteristics. A fallacy manifests, [Ralph Caplan of AIGA points out](http://www.aiga.org/the-empathetic-fallacy/), when we as designers misinterpret empathy to mean making decisions based on the observed characteristics of an individual; *truly empathic design involves understanding how that individual adjusts their behavior as a result of that characteristic, and compensating our design to meet their altered needs*.

> Empathy in design focuses on the user as a person, not just a consumer.

> Ralph Caplan, [The Empathetic Fallacy](http://www.aiga.org/the-empathetic-fallacy/), AIGA.

In his book [Designing for Emotion](http://www.abookapart.com/products/designing-for-emotion), Aarron Walter explains this as a hierarchy of our user’s needs: that a product be functional, reliable, usable, and pleasurable, in that order. In this context, we understand that an empathetic designer considers their user’s physical, mental, *and* emotional needs.

For example, the designer of a site for children may well choose a bright, whimsical tone for the pleasurable effect it would have on its young audience. However, they also may have considered that a bright palette will keep a child’s attention for longer, that a child may need more obvious visual cues in order to differentiate actionable elements, and that large, simple typography is more readable for young people.

As food for thought, consider Comic Sans. While [mocking](http://bancomicsans.com/main/) [its](http://www.comicsanscriminal.com/) [usage](http://comicsansproject.tumblr.com/) has become a bit of a game for designers, [its creator](http://www.connare.com/whycomic.htm) never intended for it to be used as body text, or really for any purpose other than to provide a friendly typeface for children’s applications. Vincent Connare, the font’s designer, realized that there was a need for a kid-friendly typeface that was not only child-like in appearance, but more readable for them as well. Its usage made children’s applications more functional and usable. Indeed, [as one teacher notes](http://stemmings.com/a-is-for-audience/),

> Comic Sans is one of the few (if not only) pre-installed typefaces readily available to the general PC user base whose lower-case ‘A’ is composed in the same manner that a child would learn to write (Comic Sans employs the latin character alpha, i.e. an ‘a’ without the hook on top). In terms of educational instruction, it’s the best available tool for the job; it literally helps to synthesize learning. In terms of its audience, *it is perfectly designed*.

## Feedback Loops and Kudos

Social trends and emotional responses are both examples of passive [feedback loops](http://uxdesign.smashingmagazine.com/2013/02/15/designing-great-feedback-loops/). A friend validates something, thus I trust it, thus my friends see that there is public support behind this thing, and so on as momentum builds. Likewise, if a website strikes an empathetic tone, it creates a more personalized experience for me, making me more likely to trust and engage with the website. As designers, we can take advantage of the strength of feedback loops and actively build them into our products.

Humans are naturally curious, and we respond strongly to positive reinforcement. Successful feedback loops take advantage of both of these facts. Some of the best examples are found in gamification, the injection of game elements such as competition, status-building, achievement, and rewards.

- [Mint](http://www.mint.com) uses positive reinforcement in the form of goals and visual graphs to make managing finances less of a chore.
- [Code Academy](http://www.codeacademy.com) and [Treehouse](http://www.treehouse.com) are two examples of companies that use goals and badges to help motivate and encourage users to attempt the daunting task of learning to code.
- [Quora](http://www.Quora.com), [Designer News](https://news.layervault.com/), and [Stack Overflow](http://stackoverflow.com/) award points to users for contributing content. They further incentivize users to ensure that their contributions are meaningful and relevant by letting other users award them points based on the quality of their contributions.

I’ll forgo a thorough discussion of the components of gamification, since there are [excellent](http://blog.kissmetrics.com/gamification-for-better-results/) [examples](http://webdesign.tutsplus.com/articles/general/the-benefits-and-pitfalls-of-gamification/) [available](http://uxdesign.smashingmagazine.com/2012/04/26/gamification-ux-users-win-lose/) [elsewhere](http://en.wikipedia.org/wiki/Gamification). What is important is that game elements add fun and personality to routine or mundane tasks, increasing engagement and stickiness.

## Trusting What’s Familiar

All of the aforementioned principles appeal to the human traits of emotion and trust. We trust the opinions of our friends. We prefer products that are personal and designed with empathy. We respond positively to game mentalities. Finally, utilizing familiar visual and interactive design elements provides the keystone to creating a credible, emotional connection between our users and our product.

[The Standford Web Credibility Project](http://credibility.stanford.edu/) found that

> …a broad range of design decisions—ranging from visual elements to information architecture to the use of advertisements—can powerfully influence whether visitors are likely to find a site credible. Like human communicators, web sites benefit (or suffer) based upon their appearance. Part of the goal of our project is to understand which design elements have an impact on credibility.

Note that they compare web sites to human communicators, emphasizing our users’ tendency to draw an emotional reaction (for better or worse) from the outward appearance of a website.

Untrustworthy design may or may not be easy to spot. One creeping issue is the prevalence of [Dark Patterns](http://darkpatterns.org/), an interface or component that is intended to trick a user into doing something. These include bait and switch tactics, hidden costs, and misdirection.

In addition to avoiding these common pitfalls, web designers can include common design patterns to help visitors feel comfortable with their interface.

- **Use consistent styling, content, and metaphors across your site and apply accessible fallbacks.** It’s easier to trust a site that is usable over one that is not. Consistent styling of links, navigation, and other interactive elements ensure that a user can interact with the site with confidence. Likewise, adhering to [code standards](http://www.w3.org/standards/) and [accessibility best practices](http://www.w3.org/standards/webdesign/accessibility) leads to providing a seamless experience for all of your users, regardless of the device that they use to access your content. A site that is well-designed from the inside-out naturally seems more professional, and therefore more credible.
- **Take advantage of common patterns**. There are things we absolutely know about how people browse the web, and designing with these standards in mind will make your site feel familiar. For example, we know that [people’s eyes track in an “F-shape” as they browse a site](http://styleguide.yahoo.com/writing/write-web/eye-tracking-where-do-readers-look-first). Knowing this, you can place navigation horizontally along the top of the page or vertically down the left of the page. We know that people expect the site logo to link to the homepage, and for content to be listed from most important to least. As responsive design continues to proliferate, users may come to expect enhanced readability without sacrificing content on a mobile device. Your design does not have to be unoriginal or overly trendy in order to adopt common visual standards.
- **Make your interfaces clear and your interactions understandable and learnable.** There are a lot of catch phrases being tossed around the design community right now: intuitive design, invisible interface, honest design, to name a few. Don’t get caught up in the hype. Interfaces can be both visible and useful if they are simple and straightforward (read [these](http://www.elasticspace.com/2013/03/no-to-no-ui) [two](http://scottberkun.com/2013/the-no-ui-debate-is-rubbish/) articles for fantastic discussions on the shortcomings of the “no UI” movement). It’s ok to introduce new interactions, even if they aren’t inherently intuitive, as long as you [teach](http://blog.timoni.org/post/45842270204/dont-be-afraid-to-teach-interactions) [them](http://52weeksofux.com/post/647891017/teach-your-users-well) effectively.

## In summary

Design for humans injects personality into the computer side of “human-computer interaction.” It uses emotional cues to inspire user behavior and make people feel comfortable and safe with our design, as though the web site or app were an extension of the person themselves. Like brands that have become household names—Kleenex, Kellogs, Google—we strive to make the usage of our products second nature. Creating an emotional connection between your site and your users leads to better retention, engagement, and trust.

---
*Originally published at [https://emilycampbell.co/writing/designing-for-humans](https://emilycampbell.co/writing/designing-for-humans) on 2017-12-29.*


# A willingness to fail

What is it about our culture that some people are so keen to fail while others aren’t? The ability to cast off from the shore with the full ability that you may sink is a trait many envy. I envy it.

In Silicon Valley, failure has become somewhat of a badge of honor. A startup founder who was willing to move on his idea, though it didn’t gain traction, still holds value to venture capitalists and the industry at large.

The reality is, it’s not the failure that’s impressive, but rather the willingness to take a risk, to expose your brainchild to the evaluation of your peers (and complete strangers) and put it all on the line. Those people who are confident enough in themselves and their skills are willing to risk having nothing in order to have it all, and more importantly, to make something that matters; something that lasts.

I struggle every day with my fear of failure. I find myself afraid to speak my mind, share my thoughts, write a personal blog post, embark on my own. Working for a [startup](http://www.hackerrank.com) has eased that. I’m better at sharing ideas that may fail. In fact, many of my ideas do fail, but that makes the ones that stick much stronger.

As designers, we have to see failure as an opportunity. We must be willing to expose ourselves to criticism and change. To know that version A-L may not be accepted, but version M will be phenomenal. By being willing to fail, we are constantly pushing ourselves forward.

---
*Originally published at [https://emilycampbell.co/writing/a-willingness-to-fail](https://emilycampbell.co/writing/a-willingness-to-fail) on 2016-12-07.*


# The empathy paradox

Empathy is [so](https://boagworld.com/usability/empathy/) [hot](http://usabilitygeek.com/empathy-and-technology-relationship-makes-good-design-better/) [right](https://www.interaction-design.org/ux-daily/123/empathic-design-is-empathy-the-ux-holy-grail) [now](http://alistapart.com/article/from-empathy-to-advocacy).

Whether you’re a user experience professional, visual designer, marketer, developer, empathy is the new skill to have. Unfortunately, like most buzz words that become jargon, the value of the word is being lost in the noise.

What really is [empathy](https://en.wikipedia.org/wiki/Empathy)? The first definition that may likely come to mind is “the ability to put yourself in someone else’s shoes.” Much like its cousins, sympathy and mindfulness, its a skill that requires emotional intelligence and awareness.

On the surface, it makes a lot of sense. By empathizing with our users (clients, colleagues, etc), we are able to create more meaningful experiences, and therefore better designed products. However, there’s a paradox to empathy: The more we *think* we know another’s needs, the less effort we make to find out what their real needs are.

A recent study out of London’s Imperial college [exposed this fact](https://hbr.org/2015/03/putting-yourself-in-the-customers-shoes-doesnt-work). After having managers describe an existing customer persona and consider the person’s thoughts and reactions, the research team asked the managers to anticipate the customer’s needs. The findings were not what you would expect. **The more empathetic managers were, the more “egocentric” they became; that is, the more likely they were to say that the customers’ preferences were the same as their own.**

The problem is, [it’s impossible to truly know how someone feels](http://harvardmagazine.com/2014/04/paradoxes-of-empathy), as much as we may think we do. When we project our emotions onto the other person, we are susceptible to thinking we understand their state of mind when we don’t. It all boils down to the five most important words to UX Designers: [you are not your user](http://52weeksofux.com/post/385981879/you-are-not-your-user).

Rather, to quote [Edith Stein](http://www.amazon.com/On-Problem-Empathy-Collected-Volume/dp/0935216111), the empathic position is one in which we know that we *are not* the other.

Once we embrace this mindset, empathy becomes a very powerful tool. Independently, we can anticipate possible areas of stress or confusion (or, joy, or excitement), then we listen to others to test our hypotheses and understand how those experiences truly make them feel.

> The only way to truly understand someone else’s perspective is to listen to them explain it.

As UX Designers, this has several implications for our process: Be careful when [dogfooding](http://blog.uxpin.com/6178/dogfood-superfood-product-development/) and remember that your reaction to the product is inherently affected by bias; recognize that many unknowns factor into someone’s impression or experience when using your product; most importantly, realize that the only way to truly understand someone else’s perspective is to [listen](http://alistapart.com/article/a-new-way-to-listen) to them express it.

---
*Originally published at [https://emilycampbell.co/writing/the-empathy-paradox](https://emilycampbell.co/writing/the-empathy-paradox) on 2015-06-14.*


# Designer: It’s what you do not what you’re called

As designers, we spend far too much time analyzing how to refer to ourselves. “Product designer, UX/UI designer, visual designer, full-stack designer, lead designer of products”…do any of these really tell you about the person behind the title? If you saw a resume with one of these titles at the top, would it shape your expectations?

My bet is, no.

For example, I have reviewed countless resumes for people heralding themselves as a UI/UX designer because they have worked with wireframes. Often times, these individuals have never once spoken to an actual user or conducted any user research.

But can you blame them? Read any job description for designers, and you’ll think you can only be hired if you can conduct research, generate content, formulate an app’s infrastructure and key interactions, test prototypes, design the whole thing using the latest trends, and then launch the product with CSS, HTML, and Javascript.

It’s no wonder designers are forced to market themselves as the full package, even if it’s just to get a foot in the door. Plus, descriptions like these contribute to imposters syndrome. Your company may be missing out on the perfect candidate because he didn’t think he fit the full description of what you were seeking.

As an industry, I think it’s time we take a step back and just embrace the title “designer”. That, in of itself, expresses so much.

A designer is a problem solver; a thinker; a doer. They are someone who can navigate complex ideas and processes and help define a clear way to manage them. And they have core strengths and weaknesses, too. But I would get more from someone saying “I am a designer that focuses on visual design and product interactions” than if they just said “I’m a UI/UXer.”

I’m sure we’re all familiar with the phrase “less is more.” The same holds true in this case.

Someone trying to describe themselves as a unicorn through their title often fails to match my expectations going in. On the other hand, I would not shortchange someone who simple called themselves “a designer.” I would feel confident that she could understand the core values behind UX research, content strategy, or interaction design, even if they weren’t familiar with the nuances and deliverables associated with that field.

> A designer is a problem solver; a thinker; a doer. They are someone who can navigate complex ideas and processes and help define a clear way to manage them.

It’s time we rid ourselves of the debates over who we are and what we do. Be a T-shaped designer, find the area you love, focus on it, cultivate your skills, but expose yourself to other areas too. Don’t be afraid to try new things, like illustrations and mobile app design.

When you’re applying for your dream job, proudly say, “I am a designer” and own it. Own your strengths, own your weaknesses, but know that you can

---
*Originally published at [https://emilycampbell.co/writing/designer-not-youre-called](https://emilycampbell.co/writing/designer-not-youre-called) on 2014-11-04.*


# User Types: The Tourists and The Explorers

User research is paramount to a well-balanced design process. It helps us create and implement interfaces that adhere to the needs and expectations of our users. But how do we adjust for different behavioral patterns amongst that group?

People tend to fall into one of two categories – thinkers and doers; cautious and impulsive; perceptive and decisive. Neither of the two is ‘better’ than the other, but they absolutely are distinct.

In a past life, I was a multi-day raft guide on the [Colorado River](https://www.google.com/search?q=cataract%20canyon&aq=f&um=1&ie=UTF-8&hl=en&tbm=isch&source=og&sa=N&tab=wi&authuser=0&ei=xM46UdifJsjOqQHSmYHABA&biw=1214&bih=775&sei=xs46UYjGLoq9qAHJ34DoAQ). Almost all of the people on my trips fell into one of these two groups. Some of my passengers wanted a controlled experience. They wanted to know where we were stopping every day, which rapids we were hitting, what we were serving for each meal, and detailed information about all aspects of the canyon itself. Other people wanted a more adventurous trip. They tended to be the ones who wanted to try and row my boat, who ran up the side hikes in front of the guide without waiting for interp, who came prepared with a swath of information that they had looked up, and who wanted to define their experience for themselves.

These two types of visitors, both unique in their own right, also shared many things in common. Namely, they still required a guide to get them down the river. They still needed assistance, but they looked for it differently. The latter group would ask which way to go and then take off to explore independently, while the former stuck close to the guide for step-by-step instruction. But they were on the same journey.

I refer to these two groups as the tourists and the explorers, respectively, and they did not leave those distinctions behind when they left the trip.

Like my passengers, your users will generally separate into these two groups. One will look for detailed walkthroughs and guidelines while the other will jump right into the interface and start playing around. However, both require a clear and intuitive interface with useful navigation and action-oriented user flows. That’s the common denominator.

*Disclaimer: For this discussion, I am going to focus on complex sites and apps, as opposed to more simple, static websites. The distinction between these two groups becomes more important when dealing with interaction-heavy interfaces. Simple websites tend to rely more on static content than complex interactions, and therefore behavioral distinctions tend not to be as dominant.*

## The Tourist

Ok, get rid of the image in your head of the middle-aged heavy weight with the fanny pack and sun visor. These users may be just as savvy and informed as their counterparts. They tend to be the people who are more motivated by short-term goals than by discovery, and they read the directions before they start to assemble an IKEA desk.

When considering this type of visitor during my planning stages, I always start with the ground level: the new user. Tourists benefit from a detailed [onboarding process](http://uxarchive.com/tasks/onboarding) that may include a walkthrough or tutorial. (Remember, though, that your other set of users may not want to sit through a lengthy tutorial, so consider making it brief and dismissible.)

Now that they have landed at a starting point, I start to address this user group (new and established) as a whole. Strong verbal and visual cues help them advance through the interface. I focus on creating goal-oriented user flows. For static content, clear pagination, breadcrumbs, and navigation provide a tangible cue to their progress, placement, and destination. For interactive active content, such as within [HackerRank](http://www.hackerrank.com), I provide progress indicators, statistics, and goalpoints to motivate progression through the site’s content. Tourists do not always need, nor want their hand held, but they do rely on [in-app education](http://52weeksofux.com/post/647891017/teach-your-users-well) to keep them engaged.

> Whether it is explicit step-by-step instructions or some subtle interface interaction that reveals more functionality, the fastest way to engage users and provide a clear path to rapid adoption is to build education right into your product from the start.

> — Joshua Porter, [52 Weeks of UX](http://52weeksofux.com/post/647891017/teach-your-users-well)

## The Explorer

This rogue individual is the ‘learn by doing’ type. She is driven by curiosity and logic – her lego creations rarely looked exactly like the image on the box. Unfortunately, the mentality of flying by the seat of ones pants can lead to frustration upon reaching a dead end. When that happens, these users will look for clear, concise navigational cues to get them back on the road to discovery, or else they will go walking.

I actually find that this is the easier of the two groups to design for. Tourists require a delicate balance of useful information that doesn’t become officious or redundant. Explorers, on the other hand, tend to rely on their own devices to learn the interface, discover its content, and trouble shoot problems.

It’s important to include useful cues to help explorers pick up new trails through the site’s content and familiarize themselves with the actions that are available to them. Joshua Brewer refers to these as [orientation, route decision, route monitoring and destination recognition](http://52weeksofux.com/post/1049250085/finding-your-way), four cues that help a user determine where they are, where they want to be, and how to get there. Strong navigation, in the form of prominent headers and footers, as well as breadcrumbs and internal links, can aid an exploratory user in finding and staying on a path.

These users tend to be curious, so encourage their discovery with suggestions and relevant actions. Popular post sections, or related content areas will drive them further into the architecture. At [HackerRank](http://www.hackerrank.com), we show detailed statistics and leaderboards to encourage users to try new challenges and refine existing code in order to advance beyond their peers. These users respond to strong [feed back loops](http://uxdesign.smashingmagazine.com/2013/02/15/designing-great-feedback-loops/), but be wary of flows that are redundant or lead to dead ends.

## The Guide

You, the designer, are your users’ guide. It’s easy to pinpoint the similarities between these two user groups. Both require useful navigation to orientate them and to direct them to the next step. Like my raft passengers, the tourist will likely follow the navigation closely in order to stay on track; the explorer will check in from time to time and use the navigation to figure out where to go next. Both sets of users require interesting content, although the tourist requires smaller, short-term chunks of information while the explorer merely requires that one chunk is just as captivating as the next in order to keep them discovering.

So how can you create interactions that appeal to both?

### In an ecommerce shop…

Create a clear and cohesive navigation and let tourists view collections and browse by specific category. Recommendations, sales, look books, and related items can grab an explorer’s attention for hours.

![](https://emilycampbell.co/wp-content/uploads/2013/03/Anthropologie-1.jpg)

*Anthropologie offers clear navigation by category to tourists, but includes plenty of hooks for explorers, such as browsing by outfit or price.*

### In applications…

Feature short term goals and communicative navigation for tourists, while helping explorers find new routes to content that relate to their behavior or interests.

 ![](https://emilycampbell.co/wp-content/uploads/2013/03/Pinterest-1.jpg)*Pinterest pulls explorers into addicting feedback loops through the content while providing equally addicting infinite navigation to tourists viewing pins in a category or board.*

### On news sites or large blogs…

Provide pagination, categorization, and archives to help tourists find clear paths through the site’s information. Catch curious explorers with strong internal links and links to featured, recommended, and popular content.

![](https://emilycampbell.co/wp-content/uploads/2013/03/NY-Times.jpg)*The New York Times provides clear categorization of their articles, giving tourists a sense of reading the paper section by section. Explorers will enjoy the popular and recommended sidebars that direct them to new articles that they may find interesting.*

Obviously not every person will fit exactly into either of these groups. However, they provide solid guidelines when considering user behaviors as you design your interface. Only by understanding what a person expects, and empathizing with what a person requires can we, as designers, create a balanced, full experience for our users.

---
*Originally published at [https://emilycampbell.co/writing/user-types-the-tourists-and-the-explorers](https://emilycampbell.co/writing/user-types-the-tourists-and-the-explorers) on 2013-03-09.*