A useful technical audit is an evidence loop: define the site and crawl scope, wait for the report to finish publishing, inspect the URLs behind a finding, fix the cause, then run a comparable crawl. An issue list by itself does not prove coverage or show that a change worked.
Define the crawl before you start
Select the property, host, and site area you want to investigate. Check path rules, depth, robots.txt handling, JavaScript rendering, the URL cap, site authorization, the estimate, and the available balance before starting.
The current Site Audit guide distinguishes verified, admin-approved sites that can use their saved full-crawl policy from unverified sites. The latter are limited to a public quick check of 100 URLs and depth 2, with robots.txt respected and rendering disabled. A blocked site cannot start a run. Verification proves control of the saved host; it does not decide whether every subdomain, login area, third-party resource, or environment belongs in scope.
For example, an audit intended for a production shop should not quietly start from a staging URL or a support subdomain. A low URL cap or shallow depth can leave important sections out. The resulting report can be accurate for its scope without describing the whole site.
Wait for the report, not just the crawl
Starting a run is not the same as completing an audit. Separate run admission, URL discovery and collection, secondary processing, and report publication. Check requested, discovered, admitted, completed, failed, and skipped counts. Link projections, findings, aggregates, and alerts may still be processing after the primary crawl work ends.
Crawl Foundry's current documentation says a run that is still processing should not be treated as a final report. A missing Site Health score means the required evidence is not ready; it is not a score of zero. A published score is versioned and retains its scope, coverage, confidence, exactness, and truncation context. Starting a new crawl does not refresh an older snapshot.
Start from a question, then choose the view
Use Overview to find areas worth investigating, Issues to trace recorded rules to URLs, Pages to create a reproducible URL set, Internal Links to inspect source-to-destination relationships, and Structured Data to review what the crawl captured. A URL dossier helps validate one case. Keep the selected site, run, and evidence date with every conclusion. The Crawl Foundry Site Audit tool gives a product-level view of URL findings and comparable crawl follow-up.
A report alone does not prove that an issue caused a visibility loss. If a different canonical appears, confirm the response, the declared canonical, and the actual target before changing a template. For a cache-related investigation, see our guide to Googlebot 304 responses.
Turn a finding into a verifiable change
Group instances by template or likely cause. Record the affected URLs, observed rule, expected change, and owner. Prioritize by the role of the pages, confidence in the evidence, the number of important URLs affected, and whether a shared cause can be fixed once. This is a practical workflow, not an official ranking formula.
After the change ships, run another crawl with comparable scope and settings. Wait for collection and publication to finish, then compare the same URLs and rules. If coverage changed, a different count may reflect the crawl scope rather than the fix.
Our broader SEO audit guide covers technical, content, and backlink review. This article focuses on the crawl-to-verification loop.
What an audit cannot promise
A crawler observes responses accessible within its scope and configuration. It does not replace Search Console, manual checks, field data, or editorial judgment. It cannot guarantee indexing, more traffic, or a particular ranking. A change is verified when the published evidence and a comparable follow-up crawl show the technical result you expected.