Do you want to go into 2027 knowing what actually changed on your site this year? A year end SEO audit is the quickest way to find out. It’s not a full technical crawl. It’s a review of the last 12 months. What did Google index? How did updates hit you? Can AI search reach and credit you? And can your data be trusted for next year’s planning?
Below are the 20 checks I use, grouped by area. Each one says what to look at, where to find it and what good looks like. Most take 10 to 30 minutes on a small or mid-sized site.
If you need the full crawl-level inspection (rendering, canonicals, hreflang, HTTPS), use my technical SEO audit checklist for that. This guide assumes the basics are covered. It asks a different question: what did 2026 do to this site?
Before You Start: Set Up a Fair Comparison
A year-end review is only as good as the date ranges you compare. Set these up once and reuse them for every check.
- Search Console: compare the last three months with the same three months of 2025. The Performance report keeps 16 months of data. So this works now, but it won’t work for much longer.
- GA4: use the same ranges. Note that explorations only reach back as far as your data retention setting allows (check 19 covers this).
- A change log: list your own launches, migrations, redesigns and plugin changes for the year. You need it to separate your changes from Google’s. I never start a review without this list.
Indexing and Crawl Checks (1 to 4)
1. The Year’s Trend in Indexed Pages
What to look at: the line of indexed versus not-indexed pages across the year, and which reasons grew.
Where: Search Console, Indexing, Pages. Click each “Why pages aren’t indexed” reason to see its own chart.
What good looks like: indexed pages rise roughly in step with what you published. Say a reason such as “Crawled – currently not indexed” grew steadily all year. That’s a quality signal on those templates, not a technical bug.
2. Sitemaps Against What You Actually Kept
What to look at: whether each sitemap still lists only live, canonical URLs, including sections you retired during the year.
Where: Search Console, Indexing, Sitemaps. Open each sitemap and check its indexed count.
What good looks like: no sitemaps for dead campaigns or old subfolders, no “Couldn’t fetch” status, and a small gap between submitted and indexed URLs.
3. Redirect and 404 Debt From This Year’s Changes
What to look at: URLs that broke during redesigns, product removals or URL changes, plus soft 404s (pages Google treats as empty or missing).
Where: the “Not found (404)” and “Soft 404” reasons in the Page indexing report. Also run a crawler on your old sitemaps or a list of last year’s top landing pages.
What good looks like: every URL that earned clicks or links in 2025 either still works or redirects in one hop to its closest match. Internal links point to final URLs, not redirects.
4. Crawl Health and Host Status
What to look at: server errors, slow response times and failed robots.txt fetches over the year.
Where: Search Console, Settings, Crawl stats. Open “Host status” and the breakdown by response code. I explain each part in my guide to the Crawl Stats report.
What good looks like: host status green, few 5xx responses, and an average response time that hasn’t crept up since a hosting or CDN change. A spike that matches a known outage date is fine. A slow rise is not.
Performance Checks (5 and 6)
5. Core Web Vitals Across the Year
What to look at: whether URL groups moved from “Good” to “Needs improvement” or “Poor”, and when.
Where: Search Console, Experience, Core Web Vitals, separately for mobile and desktop. Confirm problem pages in PageSpeed Insights field data.
What good looks like: most URLs “Good” at the 75th percentile (LCP 2.5 seconds or less, INP 200 milliseconds or less, CLS 0.1 or less). And no group that slipped after a release you can name.
6. Scripts and Tags Added During the Year
What to look at: chat widgets, A/B testing tools, consent banners, extra pixels and plugins that piled up over 12 months.
Where: your tag manager container, the plugin list, and the Network panel in Chrome DevTools on your three most important templates.
What good looks like: every script has an owner and a purpose. Anything from a finished campaign is gone. Nothing heavy loads on pages that don’t need it. In my experience, this is where a year of small additions quietly slows a site down.
Content Quality After the 2026 Updates (7 to 10)
7. Traffic Changes Mapped to Update Windows
What to look at: whether your biggest gains and losses line up with Google’s confirmed rollouts. These were the March core update (27 March to 8 April), the May core update (21 May to about 2 June), and the spam updates in March, June, August and September.
Where: Search Console Performance, with dates checked against the ranking history on the Google Search Status Dashboard.
What good looks like: you can explain each major movement. It matches an update, a change on your side, seasonality or a shift in the results page. “We don’t know” is the answer to fix before planning 2027.
8. Bylines and Author Profiles
What to look at: every author name, bio and photo on the site. In early October 2026, Google’s people-first content guidance began describing fake creator profiles as deception. Examples include AI-generated headshots, made-up names and false credentials.
Where: your CMS user list and author archive pages. My breakdown of Google’s October content guidance covers what counts.
What good looks like: every byline is a real person who wrote or reviewed the piece, with accurate credentials. Or an honest team byline.
9. AI-assisted Content and Metadata
What to look at: pages drafted with AI tools this year, and the parts people forget to check. Google’s generative AI guidance, updated on 1 October 2026, says to fact-check AI content. That includes titles, meta descriptions, structured data and alt text.
Where: a crawl export of titles, descriptions and alt text, filtered to pages published or edited in 2026.
What good looks like: no invented statistics, wrong prices or claims your business can’t back up. And no batches of near-identical pages that would look like scaled content (mass-produced pages with little value).
10. Pages That Earned Nothing All Year
What to look at: indexed pages with zero or near-zero clicks across 12 months, and pairs of pages competing for the same query.
Where: export the Pages tab of Search Console for the full year and join it to your crawl or sitemap list. Pages missing from the export had no impressions at all.
What good looks like: a short list of pages to improve, merge, redirect or remove, with a decision next to each one. You don’t need to act on all of them in December. You need the list.
AI Search Readiness (11 to 14)
11. Your robots.txt Decision for AI Crawlers
What to look at: which AI user agents (crawler names) your robots.txt blocks or allows, and whether that still matches what the business wants. Google also plans to support a shared AI training control in robots.txt from 2027.
Where: /robots.txt on every host, plus your CDN or firewall bot settings. These can block bots your file allows. My guide to robots.txt and AI crawlers lists the user agents.
What good looks like: a clear, chosen setting for each crawler. A common setup blocks training crawlers such as GPTBot and CCBot, while allowing search crawlers such as OAI-SearchBot and PerplexityBot. Googlebot stays allowed, because it controls access for Google Search, including AI features.
User-agent: GPTBot
Disallow: /
User-agent: OAI-SearchBot
Allow: /12. Snippet Controls and AI Feature Eligibility
What to look at: stray nosnippet, max-snippet or data-nosnippet settings added by plugins or developers.
Where: a crawl filtered on robots meta tags and X-Robots-Tag headers, then URL Inspection on key pages.
What good looks like: limits only where you meant them. Google’s guidance on AI features and your website says a page needs to be indexed and eligible for a snippet to appear in AI Overviews or AI Mode. No extra special files or markup are needed.
13. Generative AI Performance in Search Console
What to look at: which pages earn impressions in AI Overviews and AI Mode, and how that changed since the report appeared for your property.
Where: the Generative AI performance report in Search Console. Google began rolling it out to a subset of sites in June 2026, so not every property has it. My guide to Search Console’s generative AI reports explains how to read it.
What good looks like: your most important commercial and informational pages appear in the list. If AI impressions sit on a few old blog posts, that tells you where AI features already trust you.
14. Multimodal Searches
What to look at: pages that appear when people search with an image through Google Lens, Circle to Search or Chrome’s “Search this image”.
Where: Search Console Performance, Search type filter, Web: multimodal. The option arrived on 24 September 2026 and has no query data.
What good looks like: pages with original product or place photos show up, load quickly on mobile, and are the best landing page for someone who just photographed that thing. Treat the first date of data as a baseline, not a sudden gain.
Local SEO Checks (15 and 16)
15. Business Profile Facts That AI Now Answers From
What to look at: hours, special hours for the holidays, services, attributes and categories. Public Q&A is being replaced by AI answers in Maps, so these fields feed what customers are told.
Where: each Business Profile, compared field by field with your website’s contact and location pages.
What good looks like: the same name, address, phone and hours everywhere. December and January special hours already entered. No attribute that is out of date.
16. Review Requests and Replies
What to look at: review request templates, staff targets and reply status. Since April 2026, Google’s policy bans review quotas for staff and requests for specific review content, including staff names.
Where: your review tool settings, email and SMS templates, and the reviews section of each profile.
What good looks like: one neutral request sent to every customer, no incentives or scripted wording, and owner replies that actually show as published.
Analytics Checks (17 to 19)
17. The AI Assistant Channel in GA4
What to look at: sessions in GA4’s AI Assistant channel and the session sources inside it.
Where: Reports, Acquisition, Traffic acquisition, with Session default channel group as the main dimension. Then add Session source as a secondary dimension.
What good looks like: ChatGPT, Gemini, Copilot and Perplexity visible as sources, and custom channel groups that no longer double-count them. Remember that Google’s AI Overviews and AI Mode are not in this channel. They count as Organic Search.
18. ChatGPT Attribution Gaps
What to look at: the drop in tagged ChatGPT visits from about 8 to 15 October 2026, when links lost their utm_source=chatgpt.com tag. Also look at visits that still arrive with no referrer.
Where: a daily chart of the ChatGPT source against Direct for September to November. My guide to ChatGPT traffic showing as Direct walks through the diagnosis.
What good looks like: a note on those dates in your reports, a line in your year-end report saying the AI figure is a floor (a minimum), and a test showing your own redirects keep UTM parameters intact.
19. Key Events and Data Retention
What to look at: whether every key event still fires after this year’s site changes, and how long GA4 keeps event data.
Where: Admin, Data display, Events and Key events for recent counts. Then Admin, Data collection and modification, Data retention.
What good looks like: no key event that went flat after a redesign. And retention set to 14 months, not the 2-month default, so next year’s explorations can compare against this year. This is the one setting I’d change today if you haven’t already.
Link Checks (20)
20. Links Gained, Lost and Risky
What to look at: referring domains won and lost during the year, links pointing to URLs that now 404, and any paid or swapped links.
Where: Search Console, Links, top linking sites and top linked pages, plus Security and Manual actions. A third-party backlink tool helps with the gained and lost trend.
What good looks like: no manual actions, and every strongly linked page either live or redirected. Any sponsored placement is marked with rel="sponsored" or nofollow. Unlinked brand mentions from this year’s coverage go on your outreach list.
How to Turn the Audit Into a 2027 Plan
Twenty checks will produce a long list. Sort it before anyone starts fixing. This is the order I follow:
- Fix anything that blocks access first: indexing drops, robots.txt mistakes, broken redirects, stray snippet controls.
- Then fix trust and policy risks: fake bylines, unchecked AI text and review requests that break the rules.
- Then fix measurement: retention, key events and AI attribution, so next year’s results are readable.
- Leave growth work for January: content improvements, merges and link campaigns, planned with the evidence you just gathered.
Record each finding with an owner and a date. A year end SEO audit that ends as a spreadsheet nobody opens again was a wasted afternoon.
Conclusion
A good year end SEO audit answers three questions. Can Google and AI tools reach your best pages? Did this year’s updates and guidance changes affect you? Can you trust the numbers you’ll report in 2027? The 20 checks above cover all three without redoing a full technical crawl.
My advice: block out a day before the holidays for your year end SEO audit. Start with checks 1, 7 and 19, because they depend on data that won’t stay available forever. Then take the rest one section at a time.
