Google Search Console SEO Audit Checklist: 62 Checks in the Order That Matters

How to work through this audit

This audit runs top to bottom in a fixed order. Account checks come first, because a wrong property hides everything else, followed by penalties, crawling and indexing, page experience, traffic, links and the housekeeping that protects your history. One pass takes an afternoon for a small site. Plan about four hours for a site under a thousand pages and a full day for a store with tens of thousands of URLs, because the indexing checks scale with the number of reasons Google lists and URL inspections are rationed per property each day.

Before you start, make sure you have owner or full-user access, since some settings are visible only to those roles, and note any recent site changes. Then open the data anomalies page, where Google lists its own known reporting errors ^[1]. A drop caused by Google's logging is not your problem. Starting there saves you from chasing a traffic loss that never happened on your site.

Account and access: is the right property set up?

The audit is only as complete as the property you audit. Google calls mixing up http and https properties "probably the most common reason" for missing traffic. A typical case is a site that moved to https years ago and still has only the old http property, which shows almost no traffic and convinces the owner that organic search collapsed when nothing changed at all. These six checks take minutes.

• Property type. Audit a Domain property where possible; it covers every subdomain and protocol, and Google generally recommends it over URL-prefix properties.
• Every host covered. With URL-prefix properties, http and https need separate properties, so list every host the site uses and confirm each one is covered.
• Lasting verification. Search Console periodically rechecks the verification token, and a redesign can delete an HTML tag. DNS verification survives redesigns.
• Owners and users. Remove former staff and agencies, keep at least two verified owners from the company, and delete removed owners' tokens too.
• Email alerts. Preferences are set per user, not per property, so confirm at least one owner receives security and manual action emails.
• Analytics association. An Analytics property links to only one Search Console property at a time, and anyone with access to it may see Search Console data.

Manual actions and security: the checks that outrank everything

Two reports can wipe out results no matter how good your titles or speed are. If Manual actions lists anything, some or all of the site will not be shown in results ^[2]. Fix every listed page, then use Request Review, because partial fixes do not earn partial recovery. Reconsideration takes several days or weeks, and longer for link-related cases. A store hit two weeks before a seasonal peak can lose that whole peak. That is why these come before everything else.

• Manual actions. The report should say no issues detected.
• Security issues. Google treats this report as the source of truth for hacked content, malware and social engineering. Its sample URLs are not necessarily complete, and a review after cleaning can take a few days to a few weeks.

Sitemaps, robots.txt and crawl health

Next comes the plumbing that tells Google what to crawl and whether it can reach your server. Sizing is the usual trap: a store with 180,000 product URLs needs at least four sitemaps under the 50,000-URL limit, listed in one sitemap index. Server health is the quieter one. Repeated DNS issues make Googlebot crawl less, so a cheap DNS provider that fails a few times a week can slow how fast new pages reach the index. Google says sites under a thousand pages rarely need Crawl Stats.

• Sitemap status. Each submitted sitemap should show Success, not Has errors or Couldn't fetch ^[3]; compare Discovered pages with the count you expect.
• Sitemap limits. Up to 50,000 URLs and 50MB uncompressed per file, absolute URLs only, and no redirecting URLs. Sitemaps are not read while a manual action is unresolved.
• robots.txt fetched. The report covers up to 20 hosts. A missing file is fine, but an unreachable one stops crawling for the first 12 hours.
• Crawl Stats host status. It covers the last 90 days, and DNS failures over 5% of requests in a day count as an issue.
• Crawl response codes. Do not return 503 or 429 for more than two or three days, and turn permanent 302s into 301s.

Page indexing: read every status, not just the total

Most findings come from Page indexing, so read the trend before the reasons. Google wants a gradually increasing count of indexed pages as your site grows ^[4]. Sudden drops can mean robots.txt blocks, noindex or a login wall, while sudden spikes in not-indexed pages often mean parameter duplicates. Then work through each status Google lists. Our guide to Search Console page indexing errors covers each one in more depth.

Errors worth fixing

• Not found (404). Fix only 404s you link to yourself or list in a sitemap, and give moved pages a 301. Most 404s are harmless.
• Soft 404. A page says not found but returns a success code. Return a real 404, or fix thin pages that look empty to Google.
• Server errors (5xx). Firewalls and DoS protection can block Googlebot because it requests more than people do. Errors can be transient, so retest live.
• Redirect errors. Chains that are too long, loops and empty URLs. Shorten chains to one hop and point internal links at final URLs.

Google currently treats a 410 the same way as a 404 and drops 404 errors from the report after about a month. A large but stable 404 count from deleted products is therefore normal. A sudden spike after a release is different, because it usually means broken internal links. For soft 404s, run a live inspection and click View tested page to see what Google rendered before you change anything.

Blocks and noindex

• Blocked by robots.txt. There is "a very small chance that the page might still be indexed", so hide pages with noindex instead.
• Excluded by noindex tag. Templates often add noindex to whole sections by mistake. One wrong setting can remove a category.
• Indexed, though blocked by robots.txt. Google indexed these from links without crawling them, so they show no useful description.

The common mistake is blocking a section in robots.txt and adding noindex at the same time. It fails. Google cannot crawl the page to see the noindex, so the URL may stay indexed from links with no description for months. If a page should rank, simply remove the rule. Once a wrongly applied noindex is gone, you can click Request Indexing to ask Google to try again.

Quality and duplicate buckets

• Crawled, currently not indexed. Improve the content or merge thin pages. Resubmitting does nothing.
• Discovered, currently not indexed. Crawling was delayed to avoid overloading the site; large counts on a small site point to weak internal links.
• Duplicate without user-selected canonical. Not an error, but mark a canonical if Google chose the wrong version.
• Google chose a different canonical. If your declared canonical is not similar to the page, Google will never choose it.
• Alternate page with proper canonical tag. This is fine. Check only that the targets are right.
• Page with redirect. Expected after migrations; remove redirecting URLs from sitemaps and internal links.

Quality is the lever here. Tag pages, thin category filters and near-duplicate product variants fill the not-indexed bucket on many sites, and pruning them often lifts the indexing rate of the pages that remain. For duplicates, parameter URLs such as ?color=green are the classic source, fixed with a canonical on every variant pointing to the main page, plus internal links that use the clean URL.

URL Inspection on the pages that earn money

The reports show patterns, and URL Inspection shows single pages. There is a daily inspection limit per property, so spend it where revenue lives. Google strongly recommends that the home page is indexed, because from there it should find everything else. If its last crawl date is older than a few weeks on an active site, something is slowing Google down, and the live test plus Crawl Stats usually reveal whether it is robots.txt, DNS or slow servers. Our URL Inspection tool guide walks through every field.

• Home page. Check the verdict, the canonical and the last crawl date.
• Top money pages. Inspect the ten pages that earn the most for index status, canonical and sitemap.
• Rendered HTML. Run Test live URL, then View tested page, and compare it with what visitors see. Missing text or links point to JavaScript problems.

Page experience and enhancements

These reports judge how pages perform for real visitors and whether they qualify for rich results. The rule to remember is that a Core Web Vitals URL group takes its worst metric as its status. A group with excellent LCP and INP but a CLS of 0.3 still shows as Poor, so fixing one layout shift on a shared template can move hundreds of URLs to Good in one release. Skip the reports that do not apply to your site.

• Core Web Vitals, mobile. Good means LCP under 2.5 seconds, INP under 200 milliseconds and CLS under 0.1. Fix Poor groups first.
• Core Web Vitals, desktop. Repeat it; values reflect 75% of real page requests over 28 days.
• HTTPS report. An invalid certificate typically affects an entire site. Mark HTTPS versions as canonical.
• Rich result errors. Numbers count items, not pages, and template errors fix many pages at once. Validation can take two weeks or more.
• Merchant listings. Stores should open Shopping reports and confirm the shipping and returns setting.
• Video indexing. Google indexes one video per page, and only videos on a watch page qualify.

Traffic: what the Performance report tells you

Set the date to the last 16 months, as Google suggests for traffic drops, and compare complete weeks only. One warning matters: impressions were misreported from May 13, 2025 until April 27, 2026 while clicks were not, so click trends are the safer read across that window. CTR work often beats ranking work. A page with 20,000 impressions at a 1% CTR earns 200 clicks, and a clearer title that lifts CTR to 3% earns 600 at the same position. Our Performance report guide explains every filter.

• Trend. Use Weekly granularity to smooth weekends, and mark known site changes.
• Falling pages. Compare the last three months with the previous period and sort by click Difference.
• Page-two queries. Positions roughly 8 to 20 are close to page one; improve the matching page, not a new one.
• Low CTR pages. High impressions with low CTR mean other results look more compelling. Rewrite titles and descriptions.
• Branded share. The Branded filter has data from March 11, 2025. A site living on branded clicks is fragile.
• Device split. A large CTR gap between mobile and desktop points to layout or snippet problems.
• Country split. Unexpected countries can reveal translation or hreflang issues.
• Search appearance. FAQ rich results stopped on May 7, 2026. Do not chase retired features.

AI Overviews, Discover and News

• Generative AI performance. Live for all sites since August 31, 2026, it shows impressions in AI Overviews and AI Mode but no clicks. Compare it with the Web clicks of the same pages.
• Generative AI control. Include is the default; confirm the setting is what the business wants and document who chose it.
• Discover. Google calls it less predictable than Search, and logging errors hit it several times in 2026.
• News. Impressions count only when the link is scrolled into view.

Links, anchors and the disavow file

The Links report is a sample, limited to 1,000 rows, so read it for patterns rather than totals. If the home page holds most external links while category and product pages hold almost none, internal linking from the home page becomes the main way to pass that strength on. That is cheaper than any link-building campaign. Disavow is the opposite case: a file can hold 100,000 lines and 2MB, but a large file is a warning sign rather than an achievement, because Google says most sites never need the tool.

• Top linked pages. Money pages should earn links too. Redirect linked pages that return 404.
• Top linking sites. Google typically ignores links from known spam sites.
• Internal links. Important pages with few internal links are hard to crawl.
• Anchor text. Exact-match commercial anchors in bulk deserve a closer look.
• Disavow file. Uploading a new list replaces the old one. Delete entries you cannot justify.

Removals, migrations and housekeeping

These checks rarely find disasters, but when they do, nothing else explains the symptom. After a domain move, keep both properties verified and the 301 redirects live for at least 180 days, and Google recommends paying for the old domain for at least a year. A lapsed domain bought by someone else can capture links and traffic that still point to it. No report will warn you once it is gone.

• Active removals. Each temporary removal lasts about six months; a forgotten one may explain a missing page.
• Outdated content requests. Requests from people who do not own the site can change your snippets.
• Change of Address. After 180 days Google treats the old and new sites as unrelated.
• Recommendations. Treat them as prompts, not a full audit.
• Annotations. Each note holds 120 characters, and notes older than 500 days are deleted.

Before you conclude: baselines, cadence and priorities

Reread the anomalies page, which records issues from the last 3 to 16 months, then export Performance, Page indexing and Links data with today's date. A baseline export takes about ten minutes and turns the next audit into a comparison. If the site matters, start the bulk export to BigQuery too. The first export arrives within 48 hours, and there is no backfill, so every month you wait is history gone for good.

• Anomalies. Year-over-year comparisons across the impressions window need a warning.
• Baseline. Report exports give up to 1,000 rows, and Links exports reach 100,000.
• BigQuery. It needs a Cloud project with billing. Keep any partition expiration at 14 days or longer.
• Cadence. Small sites can repeat quarterly with a monthly Overview look. Busy sites should run indexing checks monthly and Performance weekly.

Rank what you found by business impact. Manual actions and security issues come first, indexing problems on money pages second, and experience and CTR third. Remember the limits as well. Search Console shows only your verified sites, only Google, and only top rows of data, with no competitor data or search volume. A full SEO audit adds a crawler, a keyword tool and a backlink index. This checklist is the Google side.

The bottom line

Sixty-two checks, one free tool. Most findings come from Page indexing, URL Inspection and Performance, while the rest protect the account and the history. Start with access and the anomalies page, clear manual actions and security issues before anything else, then work through indexing status by status. Run it, fix in order, validate, and repeat. Google does the counting; your job is reading it correctly.