SEO trends 2027: seven changes Google documented in 2026

Search Console reports four metrics on a classic result and one on AI surfaces.

Most "trends for next year" posts are guesswork. This one is built the other way round.

Every item below already happened during 2026, with a dated Google document or a published study behind it. Each one then says where it points for 2027 — and says so plainly where Google has not told us.

If you want the definitions first, our SEO pillar guide covers them. If you want the planning document that turns this into a quarter-by-quarter plan, that is SEO strategy. This post stays on one question: what changed, and what to do differently.

Scope is organic search only — crawling, indexing, ranking, Search Console, and the AI surfaces sitting on top of results. Paid media and social are handled in the two companion pieces in this series, a digital marketing trends 2027 post and a social media trends 2027 post.

1. Google ran more spam updates than core updates — and they got slower

Google's own Search Status Dashboard is the only authoritative log of this. For 2026 it records four spam updates and two core updates.

The two rollout windows that are publicly documented end to end are long. The March 2026 core update started on 27 March and Google confirmed it complete on 8 April — 12 days, per Search Engine Journal.

The September 2026 spam update was the surprise. Google said on its dashboard that it "may take up to two weeks to complete", and it did: started 24 September, confirmed done 8 October, almost 14 days. Spam updates used to land in hours.

Where this points for 2027: stop diagnosing inside a rollout window. A two-week spam update means week-one data is noise. Wait for the confirmed completion date, then compare. The mechanics of what core updates actually reassess are in our core update explainer — we are not re-teaching them here.

2. Search Console started reporting AI impressions — and still no AI clicks

Google added a Generative AI performance report to Search Console during 2026, and Search Engine Land reported on 31 August 2026 that it had reached all accounts globally.

Read the Search Console Help page carefully, because the gaps matter more than the feature:

• What it shows: impressions only — how many times links to your site were shown inside a generative AI feature. Broken down by page, country, date and device.
• What it does not show: clicks, click-through rate or position. Those metrics are simply not in the report.
• Surfaces: AI Overviews and AI Mode. Google does not say whether the two are separated in the data. Discover has its own report; Search Labs experiments are excluded.
• No queries. The help page does not offer a query dimension, so you cannot see which prompts you appeared for.
• Normal limits apply: the usual 1,000-row cap, Pacific Time dates, and recent data that can still move within hours.

Where this points for 2027: treat AI impressions as a visibility signal, not a traffic metric. Pair the report with your landing-page data in Search Console and accept that the join is imperfect. Anyone selling you a precise "AI traffic" number is modelling, not measuring.

3. Opting out of Google's AI surfaces became a single setting

The Search generative AI control sits in Search Console under Settings. Google's help page states it reached all websites worldwide as of 31 August 2026. Three options: include (the default), exclude, or inherit from a parent property.

What opting out actually costs, in Google's own terms:

• Your links and content are removed from AI Overviews, AI Mode and generative AI features in Discover.
• You receive no traffic and no impressions from those features.
• Google says the control "isn't used as a ranking or inclusion signal affecting other parts of Search" — so it is not a ranking penalty.
• It is not an AI-training control. That is Google-Extended, a separate thing.

There is a partial lever too. Google's robots meta tag documentation (last updated 24 March 2026) now states that nosnippet stops content being used as direct input for AI Overviews and AI Mode, and that max-snippet limits apply to them as well.

Where this points for 2027: this is a business decision, not a technical one. For most Indian brands selling something, the answer is stay in — an impression you can be cited in beats an impression you are absent from. For sites whose product is the content, the maths is genuinely different. Either way, decide it on purpose. Background in our AI search guide.

4. The click cost of AI Overviews finally got measured with a real experiment

Until 2026 most AI Overviews traffic claims were correlation. Then Saharsh Agarwal and Ananya Sen ran a randomised field experiment, reported by Search Engine Journal on 27 April 2026. A Chrome extension randomly hid AI Overviews for some participants.

• 1,065 participants, two weeks each, January–February 2026.
• Outbound clicks per search: 0.38 with AI Overviews shown, 0.61 with them hidden — a 38% reduction in clicks out to websites.
• Zero-click searches: 72% with AI Overviews, 54% without.
• Reported satisfaction and information quality were close to unchanged.

Now the caveats, because they are large. It is an SSRN working paper, not peer-reviewed. The sample was US desktop Chrome users recruited through a research panel — not India, not mobile, not everyone. The AI Mode arm of the study is described as exploratory.

Where this points for 2027: the direction is credible and the exact number is not yours. Measure your own queries. Our notes on zero-click search and earning AI Overview citations go deeper, and this explainer covers how the surface assembles answers.

5. Google wrote down what AI search does not require

This is the most useful document of 2026 and the least discussed. Google's AI optimization guide (last updated 10 July 2026) says in writing:

• "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search."
• On llms.txt specifically: "Google Search ignores them." Keeping one for other services is fine, but it is not a Search requirement.
• "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add."
• No requirement to chunk content into tiny pieces, and no need to write differently for AI.
• The actual eligibility bar: the page must be indexed and eligible to show with a snippet, and the site must not be excluded via the generative AI control.

The guide also calls optimising for generative AI search "still SEO". Separately, Google's changelog entry of 15 May 2026 confirmed its spam policies apply to generative AI responses in Search too.

Where this points for 2027: budget for indexing, snippet eligibility and genuinely useful pages — not for an "AEO file". If a vendor sells you llms.txt as a Google requirement, they are wrong, and we explain why in our llms.txt breakdown.

6. Structured data: one feature retired, several quietly expanded

Google's documentation changelog is the record here. The headline loss was FAQ.

Where this points for 2027: schema is still worth doing, just for narrower reasons — eligibility for features that still exist, and clearer entity signals. Pick markup by the SERP feature you can actually win. Our schema markup guide and FAQ schema post cover the current position.

What we changed on this site when FAQ rich results went away

First-hand, from our own site: digitalhangover.in runs on WordPress with AIOSEO, which already emits BlogPosting and BreadcrumbList schema on every post. For a while our writers were also hand-adding JSON-LD into page bodies, FAQPage included.

That created duplicate schema on 53 pages — two competing blocks describing the same page. We had to go back and fix them.

Our standing policy now, and it has not changed since:

We are not claiming a traffic outcome from this — we have not isolated one, and we will not invent one. The gain was correctness and a shorter QA checklist. Related reading: technical SEO.

7. Crawling documentation got more specific — read it before you blame crawl budget

Two quiet 2026 updates are worth knowing.

• File size limits, 3 February 2026. Google moved the default limits to its general crawler docs and gave Googlebot its own figures: the first 2MB of a supported file type, and the first 64MB of a PDF, applied to uncompressed data. The general crawler page still lists a 15MB default. Google presented this as a documentation clarification and did not announce a behaviour change.
• Robots meta tag placement, 24 March 2026. Google documented long-standing behaviour: it "doesn't enforce placement of meta robots in the HTML head and will respect robots meta tags in the body section of an HTML document as well."

Where this points for 2027: bloated HTML is a real risk if your meaningful content sits late in a very large document. And a stray noindex in the body counts — so audit for it. Groundwork in our crawl budget guide and robots.txt guide; for AI bots specifically, see AI crawlers and robots.txt.

What a 2027 SEO plan should look like because of all this

Put together, 2026 pushed organic search in one consistent direction: fewer shortcuts, slower feedback, and measurement that is honest about what it cannot see.

So a 2027 plan that holds up looks like this. Longer diagnostic windows, because rollouts are longer. AI impressions tracked as visibility, with clicks measured where clicks actually exist. A deliberate, documented decision on the generative AI control. Schema chosen for features that still exist. And no budget at all for files Google has said it ignores.

None of that is a prediction. It is just what the 2026 record already supports.