When to run this checklist
Run it on a fixed rotation, usually once a quarter, so that the findings from one run become the baseline for the next. Run it after any migration, redesign or platform change, alongside the website migration checklist, which owns the before-and-during steps while this list owns the state of the site afterwards; the checklists library describes how the lists hand off to one another. Run it when organic traffic drops without an obvious cause, when a new template ships, and before a large content push, because publishing a hundred pages onto a site with a canonical problem multiplies the problem by a hundred.
Scope the run before starting. On a small site the whole list applies to the whole site. On a large one, audit by template and section, and treat each pass through the list as covering one template; a single run that tries to cover a million URLs as one list produces a list, not findings. The international group only applies to sites with more than one language or country version, and can be skipped with a note otherwise.
Before you start
Have the tools and the data the tutorial describes: a site crawler configured to render JavaScript if the site depends on it, Search Console access with the indexing, performance, Core Web Vitals and crawl stats reports exported, a sample of server logs filtered to search engine crawlers, and the analytics property. Have the list of pages that matter, because the audit judges every finding by its effect on those pages rather than by raw counts, and have the previous audit's baseline if one exists.
Read the crawling and indexing guide first if the terms in the first two groups are unfamiliar. The checklist assumes you know what a Canonical TagTells search engines the preferred URL. does, what Crawl BudgetHow much a search engine crawls a site. means on a site large enough for it to matter, and how an indexing report's categories differ. It does not re-explain them; that is the guide's job.
The checklist
Work through the groups in the order they appear, which is the tutorial's order from fatal to cosmetic: a page that cannot be crawled makes its speed irrelevant. Each item is a check with a pass-or-fail answer for the template or section in scope; record findings against the pages they affect rather than as site totals. Ticks are saved in your browser only, so the findings themselves go into the backlog that the last group asks for.
Crawling
Indexing
Architecture
Performance
Rendering
Structured data
International
Monitoring
If an item fails
Crawling and indexing failures are fixed first, and the crawling and indexing guide is the reference for all of them: robots rules, sitemaps, canonicals, indexing categories and the difference between a page Google chose not to index and one it was told not to. A canonical pointing to the wrong URL on a template is a single fix with a site-wide effect, which is why it is checked per template rather than per page. Architecture failures, including orphans, deep pages, faceted navigation and redirect chains, are covered by the site architecture guide; redirect chainsA permanent server redirect telling browsers and search engines a URL has moved for good. in particular are usually leftovers from an earlier migration and are worth fixing even when they technically resolve.
Performance failures are worked through with the Core Web Vitals guide, which explains how each Core Web VitalsGoogle's user experience metrics. metric is measured, why field data by template is the right view, and which fixes move many URLs at once. Rendering failures go to the JavaScript SEO guide: if the rendered page differs from the source, or content only appears after a click, the crawler may be indexing a different page from the one users see. Structured data errors are fixed against the structured data guide, and markup that describes content the page does not show is removed, not corrected.
International failures are usually mechanical. A missing return link or a malformed language code makes the whole hreflangAnnotation for language/region versions. set invalid for that pair of pages; the glossary entry on hreflang states the rules briefly, and the site migrations guide covers the case where the language versions are changing URL. A monitoring failure means the audit will have to start from scratch next time; the tutorial's final step describes the backlog format, and the baseline is the first thing to record even if nothing else is fixed this quarter. Sorting that backlog on an impact-effort matrix separates the template-wide fixes that are cheap to make from the larger rebuilds that need planning and developer time.
A fashion retailer whose filters created most of its crawled URLs
A fashion retailer with about 4,000 products runs the checklist for the first time. The crawling group fails immediately: the crawler finds 212,000 URLs, and the log sample shows that roughly 70 percent of crawler requests go to colour and size filter combinations that the site never intended to be indexed. The indexing group confirms the damage, with 38,000 of those filter URLs sitting in the "crawled, not currently indexed" category and new product pages taking an average of 11 days to be indexed. The fix is a canonical from every filtered view to the unfiltered category, a robots rule for the parameter pattern, and a sitemap trimmed to the 4,600 URLs that should rank. Three months later the crawled URL count is under 9,000 and new products are indexed within two days.
Common mistakes
- Auditing by URL count instead of by page importance. A thousand broken parameter URLs matter less than one category page with the wrong canonical. Judge every finding by the pages it affects.
- Trusting the crawler over Search Console, or the reverse. The findings live where the three views disagree. A crawl alone misses what Google knows; Search Console alone misses what the crawler found.
- Checking vitals on the homepage only. The homepage is rarely the slowest template. Field data by template shows where the problem and the traffic overlap.
- Rendering checks on one page. A rendering problem is a template problem. Compare source and rendered output on each key template, not on the page that happened to be open.
- Fixing everything at once. The backlog exists so that developers get the highest-impact fixes first. An audit that produces an unprioritised list of a hundred items produces no fixes.
- Skipping the baseline. Without recorded figures, the next audit cannot say whether anything improved. The monitoring group is short and it is the one that makes the rest repeatable.
