My Visibility Index moved and I changed nothing
That can happen because AI answers are non-deterministic. The same prompt may return different brands, sources, wording, and positions on different runs. Providers also update their models without notice.
Use this triage. The five-point recheck threshold below is a provisional product safeguard, not a universal noise band or a statistical-significance claim:
| Pattern | What Stride can say |
|---|---|
| Small movement or too few comparable runs | Within normal variation; do not treat it as progress or decline |
| Repeated same-epoch movement across several scans | A directional visibility signal worth investigating |
| Movement after the prompt, engine, or scoring setup changed | Not comparable; Stride starts a new Visibility Index baseline |
| Several Actions overlapped in the same window | Confounded; the movement cannot be assigned to one Action |
Check the mention and own-site citation counts before the index. A diagnostic index movement is not proof that an implementation caused the change.
Why did my comparison trend restart?
Stride connects Visibility Index readings only when the score algorithm and the normalized prompt-text × engine work set match. A change to either starts a new baseline. Strict evidence for Actions and proof uses a separate, stricter measurement contract, so competitor or other evidence-contract changes do not by themselves reset the Index trend. Historical readings stay frozen under the formula that produced them; Stride does not recalculate old scans to manufacture a continuous line.
My scan seems stuck
Scans query live AI engines and run asynchronously, so a few minutes is normal. Longer delays can come from rate limits or provider outages. Failed provider slots are shown as not measured and are excluded from mention and citation denominators rather than counted as negative answers.
If a scan remains incomplete, email support with the site and approximate start time. Only usable completed full scans update current score, trends, Actions, and proof. Priority, incomplete, degraded, invalid, or otherwise untrusted runs can remain visible for diagnosis, but do not promote proof.
A competitor appears and I don't
That is an observed answer gap. Open the exact prompt evidence before choosing an Action. Repeated comparable scans may also show third-party sources that recur in those answers; those are contextual source opportunities, not proof that a source cited the competitor.
My brand is mentioned but not cited
The engine recognizes the brand but did not link to your site. Review which other sources informed the answer, then strengthen first-party evidence, entity clarity, and source-worthiness for that exact prompt.
Recent work is not reflected yet
AI engines need time to crawl, index, or otherwise incorporate changes. The delay varies by engine and action type and may take days or weeks. Each completed Action shows its expected leading signal, waiting period, and next measurement date. During that period, the proof timeline says settling or awaiting comparable evidence rather than declaring the work ineffective. Provider failure, insufficient evidence, an epoch mismatch, confounding, integration unavailability, or source deletion remains current evidence health; none is converted into a zero, absence, or proof of impact.