Establish what is happening
Capture the prefix and list
Record the exact keystrokes, suggestion order, labels, type, locale, and available result context. Include the first suggestion selected and a prefix that behaves well. Do not judge a list from its visual polish without the typed state.
Compare usefulness
Ask whether each suggestion completes the phrase, names a product, offers a recent search, or promotes a category. Compare accepted, ignored, and edited suggestions. Keep popularity as one signal, not proof that a suggestion matches the current job.
Set the noise boundary
Classify entries as relevant, useful but differently scoped, stale, or unrelated. Note whether grouping or a label would make a useful non-query suggestion understandable. Preserve the prefix and list size so the next review can replay it.
What to carry forward
The triage output names the prefix, suggestion types, ordering issue, and expected user job. Stop when relevant, scoped, and unrelated suggestions are distinguishable. If usefulness cannot be inferred from selection behavior, keep the list provisional and request a targeted example.
Technical background: W3C guidance.
Keep the decision with the work.
Use a Work Item in Aglet to record the problem, the evidence you have, and the next decision. Add an owner and priority, then keep updates in the discussion so the next person can follow the reasoning.
Create an account See the product workflow