Begin with revenue paths, not a defect counter
List the pages that must work for acquisition: priority services, regional coverage, proof, insights and the enquiry journey. Audit those paths first, then inspect site-wide patterns. One blocked commercial template matters more than fifty harmless notices on private utility URLs.
For each issue, record the affected URL pattern, user consequence, search consequence, evidence, owner and acceptance check. A crawler label is a lead for investigation, not automatic permission to change production.
Verify response and indexability states
Confirm the preferred URL returns the intended status without a redirect chain. Separate valid indexable pages, deliberate redirects, genuine not-found pages, private authenticated routes and temporary failures. Check robots directives in the final response and rendered HTML.
Do not use robots.txt as a substitute for privacy or a noindex decision. A blocked URL may still be known through links, while a private workspace should require authentication and remain outside public sitemaps and navigation.
Resolve canonical contradictions
Choose one preferred HTTPS host and one URL for each piece of content. Align redirects, self-canonicals, internal links and sitemap entries. Inspect parameters, trailing slashes, alternate hosts, print views and staging copies that might create competing versions.
Google treats canonical declarations as signals rather than absolute rules. Use URL Inspection to compare the declared and Google-selected canonical when evidence appears, then correct conflicting signals or improve the preferred page instead of repeatedly resubmitting it.
Test discovery through links and the sitemap
Every approved indexable page should be reachable through crawlable links from another discoverable page. Orphan pages in a sitemap are weaker than pages that belong to a clear service, proof or insight cluster.
Include only canonical, indexable URLs in the sitemap and keep modification dates truthful. A sitemap helps discovery; it does not guarantee crawling, indexing or rankings. Validate the live sitemap response and sampled URLs after every release.
Inspect what an anonymous visitor and crawler receive
Request important pages without a logged-in session and verify the main content, navigation, headings and links exist in the response or render reliably. Check that CSS, images and required scripts are accessible and that error handling does not return a successful status with an empty or misleading page.
Use text for critical service and location information. Images can support proof, but essential meaning should not depend on text embedded inside an image or on an interaction that never creates a stable URL.
Validate page purpose and visible structure
Give each indexable URL a distinct search purpose, one clear H1, useful headings, accurate metadata and copy that answers the buying question. Compare near-duplicate titles and bodies to find accidental cannibalisation before manufacturing more pages.
Structured data should describe visible reality and use stable entity identifiers. Validate the JSON, its required properties and its consistency with the page; remove unsupported awards, ratings, addresses, prices or qualifications rather than trying to enrich the markup with invented facts.
Measure performance in the customer journey
Use field data where enough traffic exists, then controlled lab tests to reproduce problems. Inspect the largest content, late layout movement and interaction delays on representative service, article and enquiry pages—not only the home page.
Prioritise bottlenecks that affect comprehension or conversion: oversized proof images, render-blocking assets, unstable calls to action, delayed forms and third-party scripts. A perfect synthetic score is not the objective if the real enquiry path remains confusing or broken.
Audit the conversion and consent boundary
Submit a real controlled enquiry and prove one stored record, the correct notification and exactly one lead event. Track form starts, email clicks and key calls to action separately so interaction volume cannot be reported as acquired leads.
Verify analytics and session-recording tools remain off before consent, honour withdrawal and mask private inputs. Technical SEO that increases visits while leaking enquiry details or losing submissions is not a successful implementation.
Check migration and historical risk
For a redesign or domain change, compare the old URL inventory with the live site, test one-to-one redirects and preserve useful content and internal references. Find broken backlinks, chains, loops, soft 404s and redirects that send unrelated pages to the home page.
Review Search Console manual actions, security issues and indexing patterns. Annotate release dates so later traffic changes can be compared with the actual technical event rather than attributed from memory.
Prioritise by impact, confidence and effort
Classify each confirmed issue by the acquisition path affected, the number of URLs, severity, evidence confidence and repair effort. Fix access, canonical and conversion failures before cosmetic metadata differences or theoretical micro-optimisations.
Where evidence is weak, run a contained test or monitor a fixed sample. Do not deploy a broad template change simply because a tool produced a high count.
Release, retest and watch the provider outcome
Use a dated release, database and source backup, explicit changed-file hashes and a tested rollback path. Retest the exact live URLs, not only the isolated test environment, and confirm queues, provider costs and private routes remain unchanged.
Then monitor crawl, index, impression and enquiry evidence over comparable periods. A green audit proves the implementation state at one point in time; it does not prove that a search engine indexed the change or that the business gained customers.