At a glance
A coverage report describes website discovery, while the document ledger describes discovered PDFs. Blocked web requests are included in failed requests; neither measure tells you how many PDFs are missing. New inventory attempts retain bounded request-level evidence so website owners can investigate recorded gaps.
Read the crawl boundary first
Check the website host, included sources, supplied URLs and saved allowances. A crawl follows reachable public links and sitemaps within that boundary. Authentication, unlinked files and JavaScript-only paths can leave gaps even when every request that was attempted succeeded.
The recorded end reason matters as much as the page count. Reaching a request, transfer or elapsed-time allowance is different from finishing reachable discovery. Coverage counters describe website requests, not PDF parser failures.
- Web pages read: successful website retrievals.
- Failed requests: discovery attempts with a recorded failure.
- Blocked requests: refused requests included in failures.
- Rule skips: addresses not fetched because of crawl rules.
Use recorded requests to investigate
New attempts retain up to 2,000 request issues per inventory across attempts. The private report lets you search and filter the recorded address, outcome, observed response code when available, attempt and time. An omitted-count indicator explains when the evidence cap was reached.
The displayed reason is a safe explanation, not the remote response body. A rule skip means the request was not fetched; it is not a server refusal. Repeated requests across attempts can appear more than once, so reconcile attempts before treating a counter as a unique-URL total.
| Recorded issue | Investigation |
|---|---|
| 403 refusal | Check anonymous access and server policy |
| 404 missing page | Check publication and stale links |
| Crawl-rule skip | Review the public crawl policy |
| Timed out | Check host reliability before another authorized attempt |
Scroll sideways to see all columns. Use arrow keys when the table is focused.
Be explicit about historical evidence
Older scans retained aggregate counts without exact failed addresses. The report cannot reconstruct those requests after the event. A current browser test is a new observation, not proof of what the crawler received during an earlier attempt.
Keep the confirmed document totals and coverage limitation together when reporting to stakeholders. Use a separate issue log for today’s investigation, then reconcile the results with a new authorized scan if needed. An incomplete crawl should never become a claimed complete website audit through a change of wording.
Sources and scope
Methods, worksheets and illustrative examples are DocAccessible editorial contributions. These references support the technical or product details described above. W3C techniques are informative guidance; they are not a certification scheme.
- DocAccessible PDF inventory
The counting workflow, report options and purchase boundary.
- RFC 9309: Robots Exclusion Protocol
How crawl rules work; robots rules are not access authorization.
- RFC 9110: HTTP Semantics
HTTP response meanings, including redirects, refusals and missing resources.
Written by the DocAccessible team under our editorial policy. Suggest a correction if a source or product detail needs updating.
See the evidence in a working example
Explore the public sample inventory report to try its table, full-screen view and missing-count reasons. The downloadable counting fixture check records what the local counter returned for seven self-authored inputs, with hashes and explicit limits.