Our Methodology
How we measure AI visibility, transparently.
DeepIntelli Research
What inputs does an AI visibility audit use?
An AI visibility audit uses a declared brand identity, canonical domain, target market and language, named competitors, buyer questions, calibrated platforms, and publicly reachable website pages. DeepIntelli records those inputs before collection so every result has a defined scope. OpenAI documents that its search products discover and cite public websites through search-oriented crawlers; the audit therefore treats crawl access and the exact collection route as evidence, not as an assumption about every model or interface.
What does Answer Presence report?
Answer Presence is the observed share of eligible audit questions whose collected AI answer names the audited brand. The output includes the question, platform, market, language, collection time, answer text, and cited sources when the platform supplies them. A brand mention is accepted only when the canonical name, registered alias, domain, or scoped product identity resolves to that brand. Ambiguous matches remain unconfirmed instead of being counted as success.
What does Share of Voice report?
Share of Voice is a like-for-like comparison of the audited brand and named competitors inside the same eligible answer set. The output lists which brands appeared in each collected answer and preserves the evidence used for the comparison. Acceptance requires one denominator: identical questions, platforms, markets, languages, and collection window for every compared brand. The result is not a general market-share estimate and does not claim coverage beyond the measured answer set.
What does Website Readiness report?
Website Readiness is an assessment of whether selected public pages can be fetched, identified, understood, and quoted as evidence. Inputs include response status, crawler rules, canonical URLs, page language and role, readable passages, authorship, review dates, source links, and structured data. The output names every selected page and every excluded page with its reason. Google recommends people-first, reliable content, while Schema.org supplies the machine-readable vocabulary used to describe page entities and relationships.
How are fixes accepted and re-measured?
A readiness fix is accepted when the same instrument that reported the gap no longer reports it on a fresh run. Crawl and schema fixes are re-fetched from the public URL; content findings are re-scored from the revised passage; Answer Presence and Share of Voice changes are re-collected only with the declared question and platform scope. A missing measurement remains unknown. Manual proof is used only when no connected instrument can observe the signal, and it never substitutes for a failed automated check.
What are the scope and limitations?
The audit measures collected, retrieval-enabled answers and the public website surface available at audit time. It does not measure sponsored placements, private account content, model training-time memorization, or an unobserved mobile interface. Platform behavior can change after collection, and an API result may differ from a consumer interface; reports therefore retain timestamps and route disclosures. Generative Engine Optimization research supports improving source clarity and authority, but it does not justify a guaranteed ranking, citation, or delivery date.