ScouTrail methodology & limitations

What we actually measure

Scans currently query the Perplexity Sonar API, a standardized, API-based observation. Support for additional engines, including OpenAI, Google AI Mode, Gemini, and Claude, is planned but not yet part of a live scan. Even a live engine's API call approximates what a grounded AI answer engine might say, but does not reproduce a personalized consumer session (e.g. a logged-in ChatGPT or Gemini app with memory and browsing history). Treat every result as directional evidence, not a literal transcript of what any specific customer saw.

Sampling, not a single score

Each question is run 3 times per engine (N=3). We report how often your brand was mentioned across those runs, not a single pass/fail score, since AI answers vary between runs even for an identical question.

"Mention order," not ranking

When we report where your brand appeared in an answer, we describe it as the mention order within that sampled answer: the position it happened to appear in that specific run. This is not an objective ranking and should not be read as one; a different run, locale, or phrasing can produce a different order.

The Visibility Snapshot is not a score

Your dashboard shows plain counts: questions with a mention, engines with a mention, answers with citations, average mention order, each traceable to the exact questions and responses behind it. There is no composite 0-100 "visibility score." ScouTrail does not measure, and does not claim to measure, an official AI ranking.

Before/after, not causation

When you mark a recommendation implemented and we rescan, we show you the before and after results. We do not claim the change caused the difference. Many factors influence what an AI answer engine returns between two points in time.