Letās start right at the core of our problem.
Teams say they care about the user experience ā they want it to be āgood,ā āintuitive,ā āsimple,ā ādelightful.ā
Okay. But hereās my question to them every time:
āWhat does good actually look like?ā
Thatās not a throwaway line. Itās the litmus test for whether youāre doing design that can be evaluated, improved, and aligned, or whether youāre designing by opinion.
UX Goals: The Experience We Want Users to Have
Letās start with UX goals.
UX goals describe the kind of experience your product should create. Not just what it does, but how people experience it.
Theyāre not about interfaces. Theyāre about outcomes in the mind, body, or emotion of the user.
Letās make this real.
If youāre building an online health insurance platform, a UX goal might be:
āUsers should feel confident that theyāve selected the right plan.ā
Thatās not a feature. Thatās a psychological state. Itās subjective ā but that doesnāt mean itās vague.
Letās try another one. If youāre designing a budgeting app for first-time users:
āUsers should not feel overwhelmed when setting up their first budget.ā
These goals drive everything else.
Where AI Can Help and Where It Canāt
Now you might be thinking: Can AI help me write UX goals?
Yes. And no.
š¤š» Where AI helps:
You can paste in messy stakeholder input, and an LLM can help you reframe it as a clear UX goal.
For example:
Input: āWe want it to be user-friendly and modernā
Output from the LLM: āUsers should be able to complete the onboarding process without assistance and feel confident using the app independently.ā
It can also help you reword qualitative research findings into actionable goals.
You drop in ten user quotes, and it might generate candidate UX goals that reflect shared needs or pain points.
š± Where AI fails:
It doesnāt know your users.
It canāt feel their frustration.
It canāt tell whether that goal is strategic for your business or important to your users.
AI can help you surface language, but not make judgments.
So: use it to brainstorm, summarize, draft. But you set the intent.
UX Metrics: How We Know the Goal Is Happening
As a follow-up, letās talk about metrics.
UX metrics are how we detect whether a goal is being met.
If the goal is āusers feel confident choosing a health plan,ā how would you know?
You might look at:
A post-task confidence rating
Number of times users change their selection
Whether they use a help guide or support chat
Time spent on comparison pages
Another example:
If your goal is āusers donāt feel overwhelmed,ā your metrics might include:
Task abandonment rate
Time spent on setup
Comments expressing confusion
This is where triangulation matters. You donāt pick just one metric. You combine several metrics, both quantitative and qualitative, to obtain a reliable signal.
How AI Can Assist with Metrics
Now: where does AI help here?
š¤š» AI can:
Suggest common metrics used for a specific goal type.
For example, if your goal involves user confidence, it might recommend the Single Ease Question (SEQ) or a Likert-scale confidence rating.Help structure surveys and questionnaires. You can prompt it:
āWrite a post-task question that measures confidence choosing between two health insurance plans.ā
And it can output usable items, sometimes with rationale.
Offer instrumentation ideas like using error logs, clickstream data, or facial sentiment coding.
š± But hereās what it canāt do:
It doesnāt know which metrics matter most to your team.
It canāt design your research for you, as it doesnāt know your constraints, timelines, or users.
And it may hallucinate; for example, it might āinventā a UX metric or cite a fake paper that doesnāt exist. You must vet every suggestion and be accountable for all of the outcomes generated at all times.
So again: AI can accelerate your thinking. Itās your sketchpad, not your strategist.
UX Targets: Defining What Success Looks Like
Now letās get to the hard truth: metrics are meaningless without targets.
A target is the threshold at which we say:
āYes, this experience is working as intended.ā
Letās go back to that health insurance example.
Metric: post-task confidence rating (scale of 1 to 7)
Target: at least 80% of users should rate their confidence as 6 or higher
That gives you a stake in the ground. Now your team can ask:
Are we meeting that threshold?
If not, whereās the breakdown?
What design or content change might help?
Or for the budgeting app:
Metric: first-session completion rate
Target: 90% of new users should complete setup without needing help or exiting
Targets also allow you to manage scope. Youāre not chasing perfection; youāre aiming for sufficient usability, confidence, and trust.
Where AI Can and Canāt Help with Targets
š¤š» AI can help you:
Suggest target ranges, based on industry averages, especially if you prompt it well.
Compare multiple examples side by side.
Phrase targets clearly and in stakeholder-ready language.
š± But it cannot:
Know whatās realistic for your specific user base.
Predict feasibility within your constraints.
Account for accessibility needs, cultural context, or technical debt.
Worst case? Youāll end up with āaspirational fictionā that nobody on the team believes in.
As we always say, AI is great atĀ helping you write a sentence.
But it cannot tell you if itās truly the right sentence to write.
So⦠What Does āGoodā Look Like to You?
Letās bring it back to where we started:
āWhat does good look like?ā
If your team canāt answer that, clearly and measurably, then youāre not designing; youāre guessing.
Let me be blunt:
You cannot improve what you refuse to define.
And if you donāt define it, someone else will ā a stakeholder with a louder voice, or worse, a metric that doesnāt reflect the user experience at all.
Thatās why this framework (goals, metrics, and targets) is not ānice to have.ā
Itās the foundation of serious UX strategy.
What Weāve Learned
Letās recap in practical terms:
UX Goals define the experience we want users to have: confidence, clarity, ease, and success.
UX Metrics help us detect those experiences through observable signals such as task success, time, sentiment, and error rate.
UX Targets tell us what āgood enoughā looks like; the thresholds we can commit to and track.
Together, they let us align design with user reality and communicate with product, engineering, leadership, and AI systems in a common language.
The Role of AI
They give you a second brain for reframing fuzzy ideas into crisp UX statements.
They surface metrics and instruments you might overlook.
They help you prototype thinking: fast, repeatable drafts, but not final answers.
Hereās what they donāt do:
They donāt understand your users.
They donāt know your market.
They donāt own your risk.
So, the best take-home message of this post would be to use AI not to replace research, strategy, or ethics, but to amplify your clarity, your planning, and your iteration.