HOW DISCOVERGRAPH WORKS
Evidence fetched directly from the submitted public website, including HTML, links, metadata and machine-readable resources.
EXTERNALEvidence returned by an independent configured service, such as PageSpeed Insights.
AI INFERENCEOptional model interpretation of a compact, sanitized evidence package. It is not treated as observed fact.
RECOMMENDEDActions derived from the evidence and product interpretation.
DiscoverGraph's deterministic audit evaluates separate dimensions such as traditional search readiness, AI retrieval readiness, entity understanding, citation readiness and agentic capability coverage. These are first-party readiness measures. They are not claims about live rankings or external visibility.
DiscoverGraph can treat clearly labeled first-party homepage sections such as About, Contact/Support, How it works, Privacy/Data handling and Security/Trust as discoverable buyer information; a separate URL is not required when the information is explicit on the page.
Representative-page discovery avoids obvious installer/archive/download links. If a page-like URL unexpectedly resolves to an attachment or binary MIME type, DiscoverGraph stops before downloading the response body. Dedicated JSON/XML/text machine-resource checks remain separate.
Product name, category, likely audience and primary conversion are inferred from weighted first-party signals such as structured data, metadata, titles, hero/summary language and prominent calls to action. Confidence can be reduced when signals conflict.
If DiscoverGraph receives a verification/challenge page instead of the real product content, it suppresses product-readiness scores and normal marketing conclusions. The report becomes diagnostic-only and the normal daily scan slot is refunded.
PageSpeed and optional AI analysis are fail-soft enrichments. A PageSpeed or AI failure does not invalidate a successful core audit. External services can time out, rate-limit or return incomplete data.
Competitive Audit runs the same audit engine for each submitted site. Each site uses one normal daily scan slot. Results compare available evidence; they do not establish market leadership or predict business outcomes.
DiscoverGraph does not claim to measure live search positions, keyword volume, prompt-level AI appearances, actual AI citations, share of voice, third-party reputation, conversion rate, revenue impact or referral analytics unless a future report explicitly identifies a verified external source for that measurement.
Methodology evolves during beta. Material changes will be reflected in the engine version and this page. Score improvements/regressions are treated as directly comparable only when the two meaningful scans use the same scanner engine version; cross-version differences may reflect methodology changes.