An SEO audit is not an 80-page report that nobody reads. It is a checklist that runs in a specific order to avoid wasting hours chasing ghosts. What follows is an operational checklist, designed to be executed from top to bottom, with the points that actually move the needle and those that usually generate false positives.
Before starting, a warning that rarely appears in guides: do not audit a site that has just migrated or been redesigned. You need at least four weeks of stable data in Search Console. Auditing during the aftermath of a deployment will make you diagnose problems that are already resolving themselves.
Block 0: audit preparation (before touching anything)
This phase is almost always skipped and is the cause of half of all incoherent reports.
1. Define the scope and type of site
An ecommerce site with 40,000 URLs is not audited the same way as a corporate website with 60 pages. In the former, crawl budget and filter facets are the core; in the latter, internal linking and duplicate titles carry more weight. Write the scope in one line before opening any tool.
2. Set the comparison period
Choose comparable 28-day windows (avoid December versus January in seasonal sectors). Search Console limits history to 16 months, so if you need more, export now.
3. Verify access and properties
Check that you have domain property (not just URL-prefix) in Search Console, read access to Google Analytics 4, and server credentials or at least access to a cPanel-type panel to review logs and headers. Without access to logs, your technical audit will be partial: say so in the report instead of glossing over it.
4. Document the initial state
Capture clicks, impressions, average CTR and average position for the last 28 days. Also save the number of indexed URLs (site: is a crude approximation; use the Pages report in Search Console). This snapshot is your baseline and prevents later discussions about what improved.
Block 1: indexing and crawling checklist
5. Indexing coverage
Review the Pages report and classify the reasons for exclusion. Prioritize two: “Crawled, currently not indexed” (a sign of weak content or cannibalization) and “Duplicate without user-selected canonical” (canonical conflict). The rest is usually expected noise.
6. Robots.txt and meta robots
Look for accidental Disallow: / and noindex inherited from staging environments. A classic mistake: blocking a URL in robots.txt that also carries noindex; Google will never see the tag and the URL may still appear in results without a description.
7. XML sitemap
Verify that the sitemap only contains canonical, indexable URLs with a 200 status. A sitemap with 301 redirects or noindex pages is a sign of neglect that consumes crawl budget. Also check that it is declared in robots.txt and submitted in Search Console.
8. HTTP status codes
Crawl the site and group the 4xx and 5xx responses. Watch out for 404s that were previously pages with traffic: if they had external links, they deserve a 301 to the closest equivalent. Intermittent 5xx errors are the most dangerous because they do not appear in a one-off crawl; review logs to detect them.
9. Redirect chains and loops
A redirect that goes through three hops loses signals and slows things down. Fix chains so that each old URL points directly to the final destination in a single hop.
10. Crawl budget (large sites)
Analyze server logs and calculate what percentage of Googlebot hits each type of URL receives. If filter or parameter URLs take more than 30%, you have an architecture problem, not a content problem. Here the solutions involve robots.txt, canonicals or nofollow internal links, and each has side effects that should be evaluated separately.
Block 2: technical performance checklist
11. Core Web Vitals by template, not by site
The Search Console report groups by “URL groups.” Audit a sample of each template (home, category, product, blog) because a template with a heavy carousel can drag down the entire group. LCP above 2.5 s on mobile is the most frequent problem; it usually comes from unsized images or render-blocking fonts.
12. JavaScript rendering
If the site is an SPA, compare the source HTML with the rendered DOM. A common mistake: critical content injected only by JavaScript that Google indexes with delay or does not index at all if there are console errors. The “Inspect URL” test with a screenshot shows you what Google actually sees.
13. Canonicals and duplicates
Check that each page has a self-referencing canonical unless it is intentionally a variant. Duplicates caused by tracking parameters or by versions with and without a trailing slash are better resolved on the server than with tags.
14. Hreflang (if there are multilingual versions)
Check that the tags are reciprocal. An hreflang pointing to a URL that does not return the link back is silently ignored.
15. Security and certificates
Valid certificate, no mixed content, and HSTS if applicable. A certificate that expires for 48 hours can cause Google to lose trust in the domain for longer than it takes to renew it.
Block 3: on-page and content checklist
16. Titles and meta descriptions
Export all titles and detect duplicates and truncations. Do not chase an exact character length: Google rewrites titles frequently. Prioritize that each title describes the page’s intent and does not repeat the brand pattern on every one.
17. Heading structure
One H1 per page and a logical hierarchy. More important than the number of H2s is that the headings reflect the subtopics a user would search for.
18. Keyword cannibalization
Cross-reference Search Console queries with the URLs receiving impressions. When two URLs split the same queries for weeks, merging is not always necessary: sometimes differentiating the intent is enough. Premature merging destroys pages that were working for different queries.
19. Thin content and orphan pages
Identify pages with fewer than three incoming internal links. Orphan pages are not discovered through navigation and depend solely on the sitemap. Linking to them from relevant content usually has more effect than rewriting them.
20. Internal linking
Check that the pages you want to rank receive links from those with the highest internal authority. The ideal click depth is three levels; beyond four, crawling suffers.
21. Images and multimedia
Modern formats (WebP/AVIF), width and height attributes to prevent layout shifts, and descriptive alt texts. Alt attributes stuffed with keywords are a burden, not an advantage.
22. Inbound link profile
Review the evolution of referring domains and detect anomalous spikes. A sudden growth of links from low-quality directories does not usually require immediate disavowal; in many cases Google already ignores them. Document the decision, but do not act out of panic.
23. Broken outbound links
Less critical than it is made out to be, but it annoys the user. Prioritize those pointing to resources you have moved yourself.
24. Unlinked mentions
Identify brands mentioned without a link in industry media. It is the link building route with the best medium-term return, although it requires editorial outreach and does not scale with automation.
Block 5: measurement and closing checklist
25. Goals and events in analytics
Verify that key events (form submission, phone click, add to cart) are recorded correctly. It is common to find duplicate conversions or events that do not fire on mobile.
26. Attribution
Check that organic traffic is not contaminated by mislabeled UTM campaigns. A labeling error can make the organic channel appear to grow when in reality it has only changed its name.
27. KPIs and alert thresholds
Define what you would consider an anomalous drop (for example, more than 20% of clicks over 14 days compared with the average of the previous four weeks). Without a threshold, any decline triggers a false alarm.
Prioritization: how to turn the list into a plan
A checklist without prioritization is a wish list. Classify each finding along two axes: impact (visitors affected × severity) and implementation cost. Indexing blocks and 5xx errors always come first. Content improvements and link building come after.
A practical criterion that saves discussions with development: present technical findings with the example URL, the screenshot of the problem and the specific solution. “Optimize Core Web Vitals” is not an actionable task; “remove script X from the product template, which blocks rendering, and defer the loading of carousel Y” is.
Limitations worth acknowledging
No audit detects everything. Crawling tools do not execute JavaScript the way Google does in all cases, so rendering problems require manual inspection. Search Console data has a two-to-three-day delay and is subject to privacy thresholds on low-volume queries. And the impact of a fix is not immediate: indexing changes can take weeks to be reflected, and content improvements compete against the rest of the SERP, not just against your previous versions.
If the site receives fewer than 500 organic clicks per month, an exhaustive technical audit is probably not the priority: the problem is usually one of intent and content architecture. Acknowledging this avoids investing weeks in fine-tuning headers while the root of the problem remains intact.