A good critique reads like a walkthrough with a spine — who, what works, what hurts, and what to do next.
A critique isn’t a list of things you dislike. It starts with who is using the screen and what they came to do, walks the task as they would, then names each snag against a known heuristic and rates how much it hurts. You call out what works too — and you always end with one concrete recommendation the team can act on.
Interactive · critique a checkout
Tap each flagged spot on this checkout screen. Tag it with the heuristic it breaks (or the one it honours) and a severity. Then order your critique below and watch the coach meter respond.
Every flagged row hides a usability point — some good, some not. Start anywhere.
Now deliver it · put the critique in spoken order
Reorder the lines into the order you’d actually speak them.
The same five moves, every time. It scales from a hallway chat to a formal review:
1. Context — who is the user and what are they trying to do?
2. Works — name one thing that is done well (earns trust)
3. Issues — walk the task; for each snag:
heuristic (why it is a problem)
severity (high / medium / low = how much it hurts)
4. Order — sort issues high -> low; lead with the costliest
5. Recommend — one concrete, owned next step
Severity, roughly:
high blocks or misleads a real task -> fix first
medium slows or frustrates, recoverable -> fix soon
low polish; small friction -> batch later| Fits | Keep in mind |
|---|---|
| Design reviews and crits | Tie every note to a heuristic, or it sounds like taste |
| Reviewing a PR’s UI or a mock | Severity is a judgement call — say why, don’t just rate |
| Whiteboard “critique this screen” interview prompts | Time-boxed: 3–4 issues beat twelve nitpicks |
Interviewer: “Critique this checkout.” You start with context — “A returning shopper on mobile trying to finish a $60 order.” You note what works: the “Step 2 of 3” indicator sets clear expectations. Then, walking the task: the total is hidden until the last step — visibility of system status, high; the card field only validates on submit — error prevention, medium; “apply” on the promo gives no confirmation — a low status gap. You land it: “Surface a running total on every step first — it’s the one change that most moves completion.” Five moves, ninety seconds, and the interviewer knows exactly how you think.
A teammate opens with: “The submit button and the cancel button look identical, so people mis-click.” Which heuristic best names it?