Quiet redesigns

The temptation, halfway through any redesign, is to make the change visible. New typeface. Bolder colour. Animation on every link. The client has paid for a redesign and the redesign should be seen to have happened.

The opposite case, less often made, is for a redesign that almost no-one notices. The spacing is finally right. The typography feels calmer because three things are doing the work that seven things were doing. A returning visitor cannot quite say what changed but reports that the site feels nicer.

This is harder work than it looks. It also tends to age better.

When loud redesigns make sense:

• The brand has materially changed. Acquisition, repositioning, new audience.
• The original was actively harmful, illegible, inaccessible, untrustworthy.
• The client genuinely wants a public statement and has the budget and reasons to support it.

When quiet ones make sense:

• The brand is fine, the website has just drifted.
• Type and spacing are doing too many different jobs.
• The site has accreted UI from successive contractors who never spoke to each other.

In the second category, a quiet redesign is usually the better commercial decision. A loud redesign is usually the better marketing decision. Knowing which the client actually needs, rather than which they have come in asking for, is most of the conversation.

The studios that consistently win awards do loud redesigns. The studios that consistently keep clients for ten years do quiet ones. Both are valid. Almost no studio I know does both well.

Two redesigns I keep coming back to as references for the quiet kind: the New York Times’s masthead and section type, which has been refined incrementally for decades, and Pentagram’s work for The Guardian over multiple cycles. Neither was loud. Both held up. The companion post on when a brand outgrows its system covers the related case where a redesign is not the right move at all.

Decide which kind of studio you are running, before the next redesign brief lands. The brief will read very differently depending on the answer.