Why a loop
A redesign is a bet. A learning system improves the odds over time.
One launch bundles many assumptions and often stops before the team learns which decision mattered. A monthly loop isolates the most important friction and retains what the evidence teaches.
01 · Diagnose
Start with the largest uncertainty.
Use funnel behavior, customer service, search, returns, recordings, and interviews to locate where customers lose confidence and why the business cannot answer the question yet.
02 · Test
Write a hypothesis that can be wrong.
Name the audience, observed friction, proposed change, expected behavior, and decision metric. If traffic cannot support a valid experiment, use structured qualitative evidence instead of pretending noise is certainty.
03 · Compound
Store wins and losses as reusable knowledge.
Successful patterns enter the design system; failed hypotheses enter the decision record. Both prevent the next team from starting at zero.