In short: before uploading, check that engines can explore the right URLs, understand their subject, choose the canonical version and access the main content. The SEO acceptance testing must focus on the domain actually published, not just on pre-production.
A technical SEO audit does not replace useful content, but it prevents avoidable flaws from making it invisible. A forgotten noindex tag, an incorrect canonical or a rendering dependent on a failed script can neutralize several weeks of editorial work.
This checklist follows the path of an engine: discover an address, receive a response, interpret the page, index it and understand its relationships with the rest of the site.
1. Check HTTP statuses and domain variants
Indexable pages respond in 200. Old moved addresses directly return a permanent redirect to the best destination. Non-existent URLs produce a real 404 or 410, and server errors are not hidden by a page that would still respond in 200.
HTTP, HTTPS, versions with or without www and case variants must converge to a single canonical version. Also check the final slashes and frequent settings to avoid inconsistent duplication.
2. Control robots.txt without giving it an excessive role
The robots.txt file indicates what robots can explore. It does not guarantee that a URL will never be indexed if it is known by other means. To remove a page from the results, use an appropriate directive on a resource accessible to the robot, then check the result.
The file must be available at the root, not block the resources necessary for rendering and declare the canonical sitemap. Read it again after each change of environment: a pre-production rule copied into production is a classic incident.
3. Generate a clean sitemap
The sitemap contains only canonical URLs, indexable and correctly responding. Drafts, private pages, redirects and errors have no place. The modification date must correspond to a significant change, not be artificially renewed with each request.
A sitemap facilitates discovery, but does not replace the internal mesh. An important page must receive HTML links from the navigation, a category page, an article or another relevant resource.
4. Review indexing directives and canonicals
For each template, inspect robots, googlebot, the header X-Robots-Tag and the canonical tag. The canonical page must usually reference itself with a correct absolute URL. A pagination, a filter or a printable version requires an explicit decision.
Do not use canonical as a substitute for a messy architecture. This is a preference signal between identical or very similar content, not a tool to redirect the meaning of a page to a different subject.
5. Check the initial HTML and rendering
The main title, useful text and important links must be accessible in the rendering that the engine receives. JavaScript applications can be indexable, but excessive dependence on client rendering complicates diagnostics and can delay content discovery.
Test the page with JavaScript disabled when it helps to understand the dependency, then inspect the rendered DOM. Check for loading screens, API errors, and hidden content behind an interaction.
Google's official best practices for AI features remind us that important content must remain available in text form and easily findable through internal links.
6. Validate titles, descriptions and semantic structure
Each page has a descriptive and distinctive title. The description honestly summarizes the content and gives a reason to consult the page without stacking keywords. Engines can adapt the display, but these fields remain important sources.
The document contains a coherent H1, then hierarchical subtitles. Links use a label that announces their destination. Buttons trigger an action; links change pages. This clarity also helps assistivel technologies.
7. Control the images
Meaningful images use a descriptive file name, appropriate dimensions, a modern format, and contextual alternative text. Decorative images have an empty alt. The dimensions are reserved to avoid layout movements.
The social image and the one declared in the structured data must be representative of the page, crawlable and sufficiently detailed. The official guide to image SEO emphasizes relevance, context and useful alternatives rather than keyword stuffing.
8. Validate structured data
Choose the most accurate Schema.org type that actually corresponds to the visible content: Organization, Service, Article, BreadcrumbList or other relevant type. Properties must not advertise a rating, price, FAQ or missing availability from the page.
Test the JSON-LD syntaxically, then with the appropriate tools. Valid structured data improves understanding and can open eligibility for certain presentations; it does not guarantee an enriched result or position.
9. Measure the mobile experience and Core Web Vitals
Check page templates, not just the homepage. Long-form pages, product pages, forms and connected interfaces can present different problems. Measure in the lab to diagnose them, then use field data when available.
The Core Web Vitals guide details LCP, INP and CLS. During the acceptance testing, pay particular attention to the main image, fonts, third-party scripts, missing dimensions and lengthy JavaScript tasks.
10. Test accessibility and interactions
Browse the page on the keyboard: visible focus, logical order, opening and closing menus, validation of forms, modals and return to the trigger. Check for contrasts, accessible names and error messages.
These controls improve usage and make the content more robust. Our web accessibility guide provides a complementary method.
11. Audit links and redirects
Launch an automatic control of internal links and then manually browse priority paths. Locate redirect chains, links to pre-production, missing fragments, unsecured protocols, and missing downloads.
When redesigning, compare the list of old URLs to the final plan. The guide preserve your SEO during a redesign explains how to build this correspondence.
12. Check public technical pages
Check at least the sitemap, robots.txt, icons, social image, legal notices, privacy policy, error pages and, if any, editorial files intended for agents. They must be consistent with the canonical domain and not expose secrets.
Security headers, the TLS certificate, and content policies don't directly improve text, but they do contribute to reliable operation. A CSP error can also prevent an image, form, or structured data from working as intended.
After putting it online
Submit the sitemap in the configured search consoles, inspect some representative URLs, and monitor errors, indexing, queries, and conversions. Exploration and indexing take time and are never guaranteed, even when all good practices are respected.
Google's Search Essentials is a useful reference to distinguish technical requirements, anti-spam policies and recommended practices.
BlackStone Digital supports the SEO audit and visibility strategy. For a usable analysis, send us the domain and priority objectives rather than a simple "SEO score" request.
