Guides
SEO audit checklist
Most SEO audit checklists are 200 items nobody finishes. This one is 44 checks in four passes — technical, on-page, ecommerce and verification — with the evidence each check needs and the window that proves the fix.
By the Visdenda research team · Reviewed by the Visdenda editorial team
Published 2026-11-19 · Updated 2026-11-19 · 11 min read
What this checklist is
A technical SEO audit checklist is the part of a full audit that tests whether a crawler can reach, render and index the page, and whether the URL set is clean enough to earn rankings. It is not the whole audit: on-page intent, internal links, ecommerce templates and the verification step come after. But it is the first pass, because a perfect page behind a blocked canonical does not exist to Google.
Run this page in four passes. The technical pass is the longest; the on-page pass checks what a reader meets; the ecommerce pass only matters if the URL set is generated by templates; the verification pass is the one most checklists skip. If you only have ten minutes, run the five checks below and schedule the rest.
The 10-minute pass
- Fetch
/robots.txtand confirm the money templates are not blocked. - Fetch
/sitemap.xmland check that it returns 200 and lists canonical URLs only. - Request one URL per template and confirm 200, self-canonical and no noindex.
- Run URL Inspection on the same URL; the live test must agree with the HTML you fetched.
- Pick the page you changed last and re-check it before reading Search Console; the index lag is one to three days.
Tools for the first three: robots.txt tester, sitemap checker and page SEO check.
Technical SEO audit checklist
The technical pass has 20 checks in four groups: access, indexation, status and rendering, and speed. Each check names the evidence to collect, not just the thing to inspect. Run it on a crawl of every template, then re-run the group you changed.
Checks that pair with a tool: AI crawler access, structured data and the check explainers.
Crawl access
- robots.txt returns 200 and allows every template that should rank. Verify: Fetch /robots.txt with curl, then test one URL per template in Search Console's robots.txt report; a Disallow on /search or /cart is fine, a Disallow on /products is not.
- The sitemap returns 200, lists canonical URLs only, and excludes redirects, 404s and noindex URLs. Verify: Compare sitemap.xml with a crawl: every <loc> should return 200 and be self-canonical; our sitemap checker flags mismatches and broken links.
- No firewall, CDN or bot rule blocks Googlebot or the AI crawlers you want. Verify: Run URL Inspection's live test on one URL per template; for AI agents, check robots.txt and fetch access in the AI crawler access tool.
- HTTP to HTTPS and www to non-www resolve to one canonical host. Verify: curl -IL each variant and confirm a single 301 to the final URL, not a chain or a 200 duplicate.
- AI crawler policy is explicit: allowed, blocked, or partly allowed, with a reason. Verify: Read robots.txt and any Content-Signal or llms.txt declaration; decide per bot, then re-check after edge or WAF changes.
Indexation and duplicates
- Every page that should rank returns 200, is self-canonical, and has no noindex. Verify: Crawl the template, then spot-check in Search Console URL Inspection: the rendered canonical and robots directives must match the HTML.
- noindex is limited to pages that must stay out: cart, account, internal search, staging. Verify: Group the crawl by robots meta and X-Robots-Tag; any indexable page with noindex, or any private page without it, is a finding.
- Canonical targets return 200 and are themselves indexable. Verify: Resolve every canonical href to a status and robots directive; a canonical pointing to a redirect or a noindex page is a dead end.
- Pagination is self-canonical; page 2 is not canonicalised to page 1. Verify: Open a paginated series and read the canonical on pages 2–3; Google ignores rel=next/prev, so the canonical is the signal that counts.
- Parameter and filter URLs are canonicalised or blocked before they multiply. Verify: Crawl a few filter combinations and check the canonical, then confirm the sitemap does not list them.
- No near-duplicate clusters: same title, same description, same body across URLs. Verify: Group the crawl by title and by content hash; the largest groups are the templates to fix first.
Status codes and rendering
- Key templates return no 5xx under normal load. Verify: Check server or CDN logs for 5xx by path, then re-fetch the worst paths with curl; a transient 5xx during a deploy still counts.
- Redirects are single-hop 301s, not chains. Verify: curl -IL the old URL and count the hops; more than one 301 is a chain to collapse.
- Missing pages return 404 or 410, not a 200 soft-404 page. Verify: Request a known-missing URL and confirm the status code and the body; a 200 with 'not found' copy is a soft 404.
- Primary content and links exist in the initial HTML, not only after JavaScript. Verify: Fetch the URL with JavaScript disabled and compare the visible text and links with the rendered DOM; parity gaps are the finding.
- Lazy-loaded images expose src or srcset in the HTML. Verify: View source and search for the image URL; if it appears only in a script, the crawler may never see it.
Speed and Core Web Vitals
- LCP p75 is 2.5 seconds or less on mobile. Verify: Read field data in PageSpeed Insights or the CrUX dashboard; lab data is a diagnostic, not the p75 users experience.
- CLS p75 is 0.1 or less. Verify: Read the same field report; reserve space for images, ads and embeds that load late.
- INP p75 is 200 milliseconds or less. Verify: Read the field report and profile the slowest interaction; third-party tags are the usual cause.
- Structured data parses and matches the visible page. Verify: Run the page through the structured data tool or Rich Results Test; price, availability, author and ratings must match what a user sees.
On-page SEO audit checklist
Ten checks, split between the snippet a searcher sees and the links and content a reader meets. These are the checks that survive a template rewrite; the technical pass tells you whether the page can rank, this pass tells you whether it deserves to. The link checks pair with the internal linking audit.
Snippet and intent
- One H1 per page, and it states the page's job in the reader's words. Verify: Count H1s in the rendered DOM; two H1s or a logo H1 is a template fault, not a copy preference.
- Title is roughly 30–60 characters and puts the primary term early. Verify: Check lengths in the crawl after entity decoding; longer titles get truncated or rewritten by Google.
- Meta description is 120–160 characters and states a specific outcome, not a keyword list. Verify: Check the crawl and the SERP preview; Google rewrites descriptions that read like keyword stuffing.
- The page answers the query's main subtopic in the first screen. Verify: Search the query yourself and compare the top results' first 100 words with yours; if they answer and you preamble, rewrite the opening.
- One page owns one intent; near-duplicate pages do not compete for the same query. Verify: Search site:yourdomain.com plus the query, or group the crawl by title; consolidate the pages that target the same job.
Links and content
- Every important page is within three clicks of the home page. Verify: Read click depth from a crawl graph; pages that only appear in the sitemap are the orphans to link.
- Anchor text describes the destination, not 'click here' or 'read more'. Verify: Export the inlink anchors for a sample of pages and read the distribution; generic anchors should be the exception.
- Headings follow the hierarchy and describe the sections a reader gets. Verify: Read the H2/H3 outline from the crawl; skipped levels or heading-only styling are template faults.
- Images that carry meaning have alt text; decorative images do not. Verify: Check the alt attributes in the rendered DOM, not the CMS field.
- The page has a visible updated date and a real change, not a date-only edit. Verify: Compare the last content change with the date shown; a date stamp without a content change is noise.
Ecommerce SEO audit checklist
Only run this pass if product, category or filter templates generate the URL set. The eight checks are the ones that change with stock and facets; the full store-specific version, including the 12-check template table, is in the ecommerce SEO audit guide.
- Facet URLs are indexable only where they have demand; the rest are canonicalised or blocked. Verify: Sample a colour, size and price filter; read the canonical and confirm the sitemap excludes them.
- Variants are self-canonical only when the variant has its own demand. Verify: Check search demand per variant, then read the canonical on a few variant URLs; decide per family, not per store.
- Out-of-stock products stay 200 with availability in the product data; discontinued ones 301 or 410. Verify: Fetch one of each; the status, the page copy and the schema availability must agree.
- Category pages have unique copy and internal links, not just a product grid. Verify: Read the rendered page with the product cards removed; if nothing remains, the category has no content of its own.
- Product schema price, currency and availability match the page. Verify: Run the product URL through the structured data tool and compare each field with the rendered page.
- Manufacturer descriptions are rewritten where they are syndicated across retailers. Verify: Search a sentence in quotes; if other stores return the same text, treat it as duplicate copy.
- Internal search result pages are noindex and excluded from the sitemap. Verify: Fetch a search URL and read the robots meta; then check the sitemap.
- The sitemap does not explode with filter or sort combinations. Verify: Count sitemap URLs by pattern; a filter explosion is a crawl-budget leak, not coverage.
Free SEO audit checklist: template and print version
This page is the template: 44 checks, each with its verification step. There is no email gate and no PDF file to download; use your browser's print dialog to save the page as a PDF, or copy the checks into a sheet with four columns: URL tested, evidence, owner and the line that counts as done.
If you want the same checks against your own URL, the free page scan returns the findings it can verify deterministically, and our public sample report shows what the output looks like on a real site, including the checks we still owe.
Who runs it, and how often
- After every deploy
- Re-check the URLs that changed, plus one URL per template you touched. This is the single-page pass; it does not need a full crawl.
- Monthly
- Crawl the affected templates, diff run over run, and read index coverage. In-house teams usually own this; agencies run it per client.
- Quarterly
- Run all 44 checks, including the on-page and ecommerce passes, then review the checklist itself: a check nobody acts on should be removed or rewritten.
- Enterprise and agencies
- Keep the same checklist per property, but version the evidence: one snapshot per client per month, with the URL, the rule and the verification date. A shared report without evidence is a status update, not an audit.
How to verify a fix
A checklist that ends at "fixed" is a to-do list. The last six checks close the loop: reproduce the finding, re-check the exact URL, re-crawl the template, wait for Search Console to finalise, compare a baseline with a post-change window, and keep a control set.
- Reproduce the finding on the exact URL before changing anything. Verify: Re-fetch the URL and record the status, canonical, robots and the rule that failed; a sample is not the same URL.
- Re-check the single page after the deploy. Verify: Run the same rule on the same URL; a pass closes the finding, a fail records the verdict and leaves it open.
- Re-crawl the template and diff run over run. Verify: Compare added, fixed and regressed findings per URL; one run is a snapshot, not a trend.
- Wait for Search Console to finalise before reading the effect. Verify: GSC data lags one to three days; reading the same day mixes an incomplete window with the fix.
- Compare a 28-day baseline with a 14-day window that starts three days after the change. Verify: Use the same page-level window for the target and its controls; fewer than 100 impressions in either window is insufficient.
- Keep a control set and label the result as a comparison, not proof. Verify: Pick same-site pages with no change in the window; without controls, report directional-improved or directional-degraded, not a causal claim.
The method behind the window is on the methodology page; the workflow is in how to verify an SEO fix. The console runs the single-page re-check and the run-over-run diff; the effect window is a comparison, not proof of causation.
What this checklist does not do
- Local packs and listings. Google Business Profile, citations and review signals are a separate audit; this page does not score them.
- A backlink index. Link equity needs a third-party index; use Ahrefs, Moz or Semrush for that, then bring the page decision back here.
- Server-log ingestion. Crawler logs answer questions a fetch cannot; the log file analysis guide covers that pass separately.
- A live index check. Index status comes from Search Console URL Inspection, which is a point-in-time report; fetching a page is not the same as being indexed. Googlebot index explains the difference.
- Causal proof. A before/after window is a comparison with controls, not an experiment. Label it as such.
The same boundaries are documented on the methodology page; if a check depends on JavaScript rendering, say whether a render pass was available before you call it a pass.
Frequently asked questions
What is an SEO audit checklist?
It is a working list of checks with an evidence step for each one: what to inspect, how to verify it on a real URL, and what result counts as done. It is not a report and not a keyword list. A good checklist is short enough to run end to end, then repeat after the next deploy.
How many checks should an SEO audit have?
Enough to cover every template and failure mode you can act on, and no more. This page has 44 checks across four passes. A 200-item list usually means nobody owns the second half; if a check has no owner or no verification step, it is a wish, not a check.
Is there a free SEO audit checklist template?
Yes: this page is the template. There is no email gate and no account required. Use your browser's print dialog to save it as a PDF, or copy the checks into a sheet and add owner, due date and the URL you tested.
What is the difference between a technical SEO audit and a full SEO audit?
A technical audit is the crawl-and-indexation subset: access, status codes, canonicals, rendering, speed and structured data. A full audit adds on-page intent, internal links, content quality, ecommerce templates and the verification step. The technical section here is deliberately the thickest because it is where most ranking failures start.
Can ChatGPT or an agent run this checklist?
It can explain a check and draft a fix, but it cannot verify the result without fetching the URL. Run the checks deterministically, then give the agent the evidence: our MCP exposes audit findings, page reports and fix verification over one endpoint, scoped to the sites you approve.
How often should I run an SEO audit?
Re-check the URLs you changed after every deploy. Crawl the affected templates monthly. Run the full 44 checks quarterly, or when traffic, index coverage or a template changes. A full audit is a cadence, not a one-off project.