Methodology

Every number EchoRanked shows you comes with two pieces of context: a confidence band and a measurement surface. This page explains both, because a visibility score without them is noise.

Real answers, kept verbatim

We run your tracked prompts against the live AI engines and store every answer verbatim, with every citation it carried. The receipt you read on the dashboard is the actual answer, not a reconstruction — no score ever asks to be taken on faith.

Sampling and confidence bands

AI engines are not deterministic: the same prompt can mention your brand in one run and skip it in the next. So we never report a single run as truth. Each prompt is sampled N times per engine, and the result is reported as a share with a Wilson confidence interval. A 3-of-3 result shows as a tight high band; a 1-of-3 result shows as a wide middle band. If a number is based on a single run, it is flagged as low-confidence.

Measurement surface

Every external answer we record is tagged with how it was captured: api means the data came from the provider's official API; consumer means it was captured from the consumer-facing product itself. The two can disagree, and we never present api-surface data as if it reflected what a consumer saw. The surface tag is part of every result, dashboard, and digest.

Readiness checks

The static scan grades your site the way a language model reads it: llms.txt presence and validity, JSON-LD presence, robots.txt access for AI crawlers, answer-first content structure, pricing extractability, and JavaScript-only invisibility. Each check returns pass, warn, or fail with the evidence that produced the verdict.

This page will grow as the product does — sampling counts, engine lists, and bot rosters are published here as they change.