Autocomplete timing with human-reviewed evidence

Review autocomplete at the moment a suggestion changes

Record the changing suggestion list and the chosen destination so a timing-sensitive observation becomes a reviewable interaction issue.

  • ✓A row changes beneath the pointer
  • ✓Loading resembles an option
  • ✓Cleared text leaves old choices

Keep the changing choice visible in the handoff

Sequence instead of a still

Preserve the update that occurs between typing and choosing a suggestion.

Intent and result separated

Describe the intended row and actual destination without guessing the request implementation.

Multiple selection paths

Carry pointer, keyboard, and dismissal conditions into the follow-up review.

Review the moving choice

A customer types a synthetic product name into your application's search field, moves toward a suggestion, and sees the list update before selection. The clicked row no longer matches the item they intended. Scope the review around how query changes, loading feedback, highlighted choices, and selection interact. A static screenshot of the final suggestion list cannot explain this timing-dependent experience on its own.

Record typing and selection conditions

Use seeded suggestions and harmless queries rather than real search history. Capture the field before typing, successive query text, the suggestion update, and the selected destination. Record browser, viewport, input method, and any approved test network condition manually. Supported browser video capture or an authorized uploaded recording can preserve this sequence. Keep unrelated history suggestions and personal account information outside the capture.

Describe the visible mismatch

Review Retick's AI candidate findings against the original video, editing or rejecting drafts that miss the order of events. A concrete issue is Suggestion update changes the row under an active pointer. Annotate the relevant frames and distinguish the customer's intended choice from the actual destination. Do not attribute the behavior to request ordering without technical evidence; the finding can describe the changed row and its consequence without diagnosing implementation details.

Example acceptance criteria

Agree how the product should handle a suggestion list changing during an active choice. The selected label and resulting destination should correspond under that approved rule, and loading feedback should not masquerade as a selectable result. Verify keyboard selection and escape-to-dismiss directly. Reopening the suggestions should reflect the current query rather than an earlier input. Have the interaction owner confirm the intended behavior before handing it to engineering.

Recheck with distinct input paths

Organize pointer, keyboard, and cleared-query captures in a review cycle. After revision, repeat quick typing, deleting back to a shorter query, and dismissing the list before it settles. Export approved findings as files or create selected individual-media Trello cards. Retick does not execute typing tests, measure network races, or establish suggestion relevance. Timing and request behavior need direct instrumentation when important; the recording is evidence of the visible selection problem, not a complete diagnosis of its cause.

Capture the suggestion before it moves

Review the query-to-selection sequence and agree on stable choice behavior.