How we assess and improve AI visibility

01 · Define the scope

Agree on the market, language, products, platforms and collection channel. Select questions that reflect buyer decisions and identify brand and non-brand questions separately. Keep the question list versioned.

02 · Document a baseline

Record the question, answer, cited URLs and collection time. State the platform, model or tier when available, region and relevant settings. Official API observations and real website or app observations are separate channels with separate costs and reporting. Disclose failed or missing checks.

03 · Inspect the sources

Check what the cited pages support and where your website’s facts, explanations or technical access fall short. A page being accessible does not prove it was crawled, indexed or used in an answer. Each of those needs its own evidence.

04 · Agree and record changes

Connect recommendations to specific pages and evidence. Make only authorized changes, use accurate visible content and record what was published. We do not use hidden instructions, fabricated reviews or misleading claims to influence answers.

05 · Retest and report uncertainty

Repeat the agreed questions and document the conditions. Report changes, unchanged findings and inconclusive results. Answers can vary with platform, model, wording, location and time; a sampled comparison does not describe every answer or establish causation by itself.

Keep outcomes separate

A mention is not a citation; a citation is not a recommendation, visit or sale. We report identified AI referral visits, unattributed visits to deeper pages and users’ self-reported AI influence separately, without combining them into an estimated AI traffic total. Request counts are not visitor counts.