Design Metrics
A stakeholder says "I don't like how this looks." You have two options: meet opinion with opinion and let the loudest voice win, or answer with evidence. Measuring design gives you the second option - numbers tied to what users actually do, not what anyone thinks they should do.
Learn on the move: press play and turn your commute, walk or drive into time that grows your design career.
Listen to the full Design Metrics course on DesignPath
There's a bigger reason this matters. Finance, marketing, and sales all state their value in metrics; design often can't, so it gets seen as a source of cost and friction rather than value. Measurement closes that gap - it turns "trust me, this is better" into "task completion rose 18% and support tickets fell," a sentence the rest of the business can act on. You've set goals and learned the business numbers; this module is how you measure the design itself.
Design-Quality Metrics - Task Success & Top Tasks
Knowing you should measure design is the easy part. The real work is choosing what to measure - and resisting the pull to optimize a single number until it hurts the product elsewhere.
The Metrics Worth Tracking
Most design-quality metrics come down to a handful of task-level signals, plus a few standardized scores:
Task success rate:the share of users who complete a task. The single most telling design metric.
Task completion time:how long the task takes - falling time usually means a smoother flow.
Error frequency and recovery:how often users slip, and how well the design helps them recover.
Accuracy of submitted data:whether people can enter the right information without mistakes.
Onboarding times:time to first value, first purchase, or upgrade - how fast a new user gets somewhere.
Standardized scores:SUS (target 75+), accessibility, and web performance - covered in the next article.
You don't track all of them. You track the few that map to your product's most important tasks - which is the harder question.
The Top Tasks Method
Gerry McGovern's Top Tasks method is the disciplined way to decide which tasks matter most, so you measure what users actually care about instead of what's easy to count:
- Find the candidate tasks from user research and server logs.
- Have users vote on which tasks matter most to them.
- Standardize the task instructions so every measurement is consistent.
- Run usability tests with 20-30 participants every 8-12 weeks.
- Plot success rates and completion times, and track them across cycles.
The cadence is deliberate: every 8-12 weeks is often enough to catch change, rare enough that changes have had time to take effect.
Pro tip: The average user doesn't exist - these numbers aren't about any one person. They reveal patterns across the whole population and how they shift over time, which is exactly what a single interview can't tell you.
The Single-Metric Trap
The danger in measurement is optimizing one business number while ignoring what it costs elsewhere. The classic example: a product page highlights "Add to basket" in green and quietly auto-includes insurance. Spend per order goes up - and so do returns, complaints, and support costs, often by more than the conversion gain. The metric looks great; the business is worse off. (This is dark-pattern territory - see Ethical UX & Dark Patterns.)
A healthy metric set connects design signals to business goals and accounts for the full picture, not just the number you happen to be optimizing.
Where AI Helps
Prompt: "For [product type], here are our likely top user tasks: [list]. Recommend which design metrics to track for each, draft standardized task instructions for a Top Tasks test, and flag any single-metric optimization that could create hidden downstream costs."
AI proposes the metric set and task scripts; you validate against reality.It can map tasks to metrics and write consistent instructions, but which tasks are genuinely top comes from your users' votes and your logs, not the model's guess.
