Design review is difficult for mostly one reason: the feedback arrives in the shape of taste. “It feels a bit cramped”, “I don’t love this colour”, “it looks untidy” — none of which can be argued with or acted on. The person receiving it gets defensive and the meeting runs long.
Yet a great deal of that feedback is the consequence of breaking a rule you already agreed. Rewritten as rules, the argument becomes a check.
Four things that convert into rules
Most recurring feedback falls into these four.
Spacing. “Cramped” is usually a value that left the scale. If the scale is built on multiples of four, 13px and 27px are violations on their own terms and nobody has to judge them by eye. More common still is using scale values in the wrong relationship — when the gap between cards equals the gap between sections, the grouping stops reading. A scale must define not only which values are permitted but which must exceed which.
Colour. A colour that is not in the palette is a violation. This is the easiest check and therefore the most frequently broken one: hand-write a value under time pressure often enough and the palette quietly stops meaning anything.
Contrast. This is arithmetic, not taste. Do not let anyone decide it “looks sufficient”. The habitual offenders are pale grey supporting text and white text on the brand colour — brand colours tend to sit at mid lightness, which puts white right on the margin.
States. A state absent from the mockup will be absent from the build. Hover, visible focus, disabled, loading, empty and error: checking those six at design time removes a great deal of rework later.
Some things do not convert
Not everything belongs on a checklist, and forcing it there leaves you with a ritual. The purpose of mechanical checking is to leave time for the parts that need judgement.
The meeting agenda should follow the same split. Collect rule violations into a list before the meeting and spend the meeting on the right-hand column. Reading violations aloud one by one consumes most of the available time.
The third time a finding repeats, fix the rule
Consume review output purely as individual fixes and the next screen earns you the same findings. When an item comes up three times, the problem is the rule rather than the screen.
It is one of two things: the rule was never agreed, or the rule exists somewhere nobody can find it. The second is genuinely more common — documentation that lives away from the code does not get opened during the work, and a document nobody opens may as well not exist.
One format for writing a finding
A closing suggestion on wording: where, what, which rule it breaks, and what to change it to. With those four parts a finding is already a work instruction.
Instead of “the card grid feels cramped”, write “in the card list the gap between cards equals the gap between sections, so the grouping does not read; increase the section gap by one step”. There is nothing left for the recipient to interpret, and no room to hear it as taste.
Fixing spacing scales and palettes as rules is covered in the design tokens series, and contrast thresholds and state indicators in the accessibility series. How review and verification sit inside the actual sequence of work is on our process page, and if you need to check a site on your own, the diagnostic on the free tools page is a starting point.