Coverage and provenance
The report states URLs discovered, pages attempted, pages analyzed, links checked, crawl limits, elapsed time, warnings, and the exact stop reason.
Free bounded technical site crawl
What the crawl records
The crawl combines bounded public source requests, robots and sitemap parsing, an internal-link graph, cross-page aggregation, and optional Google performance evidence. Every result names what was observed; no pass claims that Google indexed, selected, or ranked a page.
The report states URLs discovered, pages attempted, pages analyzed, links checked, crawl limits, elapsed time, warnings, and the exact stop reason.
Every fetched URL records its final HTTP status, content type, bytes read, duration, and redirect path under bounded response limits.
The crawler records robots.txt policy, robots meta directives, X-Robots-Tag headers, sitemap conflicts, and eligibility for every analyzed page.
Each page records resolved canonicals, title, description, headings, exact duplicates, language, and conflicts across the crawled set.
The audit counts internal inlinks and outlinks, checks additional destinations, identifies broken and redirecting links, and labels potential orphans as sample-based review items.
Robots-declared and conventional sitemap locations are parsed, sitemap indexes are expanded within limits, and sitemap-only versus crawl-only URLs are reported.
JSON-LD syntax and declared types, missing image alt attributes, common analytics patterns, and common advertising patterns are recorded as source observations.
Public CrUX 75th-percentile values are shown when available, while a single mobile Lighthouse run and its findings remain clearly labeled as lab evidence.
Methodology references: Google sitemap guidance, canonical guidance, internal-link guidance, and OpenAI crawler documentation.
What it does not prove
This evidence report does not report keyword positions, Search Console impressions, local map-pack placement, backlinks, review strength, entity consistency, or whether an AI system cited the site. Those require different evidence and often authenticated first-party data.
It answers bounded technical questions: which public URLs were discovered, what each response and source declared, how crawled pages link together, where sitemap and indexability signals conflict, and which cross-page patterns need correction. It does not claim that Google rendered, indexed, selected, or ranked the reported URLs.
Compare against real Los Angeles sites
Spec collected a frozen 100-site Los Angeles service-business sample across ten categories. Its internal aggregate QA checks passed with 97 technical observations and a deterministic 20-site manual review. Unknown evidence remains outside every denominator. External editorial outreach is on hold until the selection ledger and independent manual-review gaps are closed.
83.2%
79 of 95 known observations.
82.1%
78 of 95 known observations.
45.3%
43 of 95 known observations.
100%
96 of 96 known robots checks; access still does not guarantee citation.
No. It does not query a rank-tracking database or Search Console. Ranking evidence should come from Search Console, a documented rank tracker, or a reproducible search observation.
Only when every same-origin candidate discovered within the configured depth fits inside the 25-page limit. Larger sites receive a bounded sample, and the report states that it is partial, how many URLs were discovered, and why the crawl stopped.
No. It inspects fetched source HTML and labels that limitation in the report. Client-rendered metadata, links, directives, or schema require a separate rendered-DOM comparison.
No. The tool checks technical eligibility signals such as indexability, structured data, and OAI-SearchBot access. It does not ask an assistant to generate an answer and does not claim citation or answer visibility.
OpenAI documents OAI-SearchBot as the crawler used to surface websites in ChatGPT search. Allowing it is an eligibility control, not a promise that a page will be crawled, selected, cited, or ranked.
No. Technical checks establish whether the page is accessible and declares key signals. Rankings also depend on intent satisfaction, content quality, internal and external links, entity evidence, local signals, competition, historical performance, and whether Google indexes the intended canonical page.
A site can time out, block the audit request, return an unexpected response, or omit a conventional sitemap location. The tool labels that state unknown instead of turning a failed verification into a false pass or failure.
No. Submitted query strings are rejected because preview links, sessions, and signed parameters can contain private values. Enter the public canonical page or site origin instead; discovered tracking and query variants are excluded from the HTML page queue.
The public URL, without query parameters, is sent to Spec's same-origin audit endpoint so the server can inspect robots first, crawl a bounded set of permitted same-origin pages, parse sitemap documents, check internal destinations, and request optional PageSpeed evidence. Email is requested only after the audit and only if you choose to receive the report.
Bring the results to a discovery call. Spec will separate what the automated scan proved, what still needs first-party evidence, and the order in which the work should ship.