What the baseline showed.
On 14 October 2026, a replay of the previous search logic found that “connected car” returned no providers, while “connected-car” found VectorDrive. “Pay per mile” returned no guides, while “pay-per-mile” found the existing guide. A By Miles search returned no visible record because the archive was hidden.
Search now normalises punctuation and spacing. Relevant aliases help find the existing telematics guide. An archived match is explained and can be revealed explicitly. These are reproducible interface checks, not visitor behaviour or evidence of increased conversion.
New articles and provider expansion wait for evidence of an unmet need. Ratings wait for a clear purpose and a participation and moderation process.
1. Start an optional usability review.
A review saves counters in this tab’s session storage for up to four hours, checked on use. It records fixed search topics and empty-result counts, worksheet starts, calculation successes/errors, sharing outcomes and task ratings. It saves no typed search terms, worksheet inputs, results, personal identifiers or page history. Nothing is sent automatically.
Normal browsing is not recorded. Closing the tab normally clears its session storage; some browsers restore sessions, so use Delete to remove this review explicitly. A Site revision change pauses recording.
No session recorded.
Try these tasks, then record the outcome.
- Use the directory to search “connected car”. Identify how mileage is measured and what is still unverified.
- Search “By Miles”. Find its archived status and the official closure explanation.
- Search “pay per mile” in the guides and distinguish it from limited-mileage cover.
- Choose any worksheet, load the invented example and calculate it. For the cost worksheet, review and tick the assumptions.
- Copy a page link. Confirm it contains no search text or worksheet values.
2. Compare stopped review sessions.
Import the downloaded JSON reports to see search failures and worksheet completion. Duplicate identical sessions count once. Different interface revisions remain separate. A completed worksheet means the calculation succeeded at least once in that session; repeat submissions are not extra people.
3. Inspect Search Console demand when available.
No Search Console data is connected. Choose an actual Queries CSV export and its date range. The file stays in page memory. Totals describe only the supplied rows. A missing literal match is a prompt to investigate, not proof that a new article is needed.
Keep an evidence summary.
The summary combines imported review counts and Search Console totals. It excludes raw query text. Imports are cleared when you leave this page. Download a summary if you want to retain it; no server inbox or shared evidence database is connected.
Choose the next change.
- Identify a concrete difficulty.Use a reproducible fault, failed search, observed task difficulty or a genuine query export. Keep reviewer tests separate from visitor data.
- Check the existing answer.Improve findability, wording or navigation before adding overlapping pages. Examine fixed topics marked “other” in a moderated task review; raw searches are deliberately not stored.
- Record a bounded decision.State the evidence, affected task, proposed change and how success will be assessed. Compare the same tasks after the change.
- Expand selectively.Add a provider or topic only after the user need and primary evidence are established. Do not infer popularity from this small catalogue or treat a small review sample as a reliable conversion rate.
Research and source methods · Privacy details · Review operations