Most redesigns fail long before the first pixel is drawn. They begin as a reaction — to a competitor, to a board meeting, to a vague feeling that things look tired — rather than as an answer to a clearly defined problem. When the brief is fuzzy, the outcome is fuzzy, no matter how talented the team executing it.

Start with the problem, not the palette

Before touching type or colour, we insist on naming the single change we want a visitor to experience. Faster comprehension? More trust? A clearer path to purchase? Everything downstream becomes a decision in service of that goal, and anything that does not serve it is cut without ceremony.

This is uncomfortable, because it means killing ideas that are beautiful but beside the point. But a redesign is not a portfolio piece — it is a promise to the people who use the product every day. Respecting them means being ruthless about relevance.

Measure what actually moved

The redesigns that succeed are the ones that ship with a way to know whether they worked. We define a small number of metrics up front and watch them for weeks, not hours. If the numbers do not move, the work is not finished, however handsome it looks in a case study.