SEO Trends 2026: Seven Changes Google Documented

Dominik Breitbach · Founder & Lead SEO Strategist

Dominik Breitbach founded taismo, an SEO and GEO agency from Munich, and reads Google’s changelog so clients do not have to. He tracks what Google actually documents and turns it into the handful of moves that pay off.

⏱ Reading time: 19 min
🔄 Last updated: 12 August 2026

The SEO trends of 2026 are documented rather than predicted. In the first seven months of the year Google shipped a generative AI performance report in Search Console, gave site owners a switch to leave AI features, ended the FAQ rich result on 7 May 2026, published its own guide to optimizing for generative AI, extended preferred sources into AI Overviews and AI Mode, and ran four ranking updates. One line connects all of it: Google is closing the measurement gap it opened when AI answers took over the top of the results page, and naming the optimization tactics it considers folklore. Each of the seven changes below carries a date and a source.

👉 Want the 2026 changes turned into a work list for your site? That is what our SEO services do every month.

What changed in the first half of 2026

Most trend articles for a given year get written in December and guess. This one has hindsight: The first seven months of 2026 are on the record in Google’s own changelog and status dashboard.

Seven changes matter enough to plan around. Google started reporting how often your pages appear inside AI answers, offered a switch to leave them, published a guide naming five popular optimization tactics it does not use, retired a rich result that ran for seven years, pushed preferred sources into AI Overviews and AI Mode, ran four ranking updates, and shipped search agents at I/O. For two years the honest answer to “how visible are we in AI Overviews” was a third party estimate. Since June there is a first party number.

Trend 1: Google now measures AI visibility, and lets you leave

On 3 June 2026 Google announced generative AI performance reports in Search Console, with separate views for Search and Discover. The report counts how often URLs from your site appeared in generative AI features, and which ones, split by page, country, device and date. Google is rolling it out to a subset of properties first, so an empty menu entry is not proof that your site is invisible.

The metric set is the part worth reading twice. The report covers impressions in AI Overviews and AI Mode, but not clicks, click through rate or a position inside the answer. Data from AI Mode also counts toward the regular Search performance report now, so year over year comparisons in Google Search Console mix two surfaces. The scale explains the urgency: Google put AI Overviews at over 2.5 billion monthly active users and AI Mode at over one billion, and said it had increased the number of inline links inside AI responses.

The second half of the announcement is the opt out. Search Console is getting a toggle that removes a site from generative AI Search features, tested with UK site owners first, and Google is explicit about the cost: Sites that opt out “will not receive traffic or impressions from our generative AI features.” Leaving buys no compensating visibility elsewhere, so the only sound reason is a licensing position, never a traffic tactic.

What this changes in your quarter: Set up the export before you set up the expectations. Four dimensions are on offer, pages, countries, dates and devices, under the same thousand row limit as the regular performance report, so a large site needs filtered exports rather than one download. Two definitions then decide whether the numbers mean anything. An impression counts once per feature even when several of your pages turn up in the same answer, and data from Search Labs experiments is excluded outright. Anyone holding this report next to a third party AI visibility tool is holding a reach figure next to an estimate of citations.

Trend 2: Google answered the GEO debate itself

On 15 May 2026 Google Search Central published a guide called “Optimizing your website for generative AI features on Google Search”, updated again on 10 July 2026. Two years of agency debate about AEO, GEO and LLMO finally have an official counterpart. It states that “the best practices for SEO continue to be relevant because our generative AI features on Google Search are rooted in our core Search ranking and quality systems”, and reframes the vocabulary: Optimizing for generative AI search is optimizing for the search experience, and thus still SEO.

More useful than the framing is the list of tactics Google calls unnecessary. Google Search itself does not use llms.txt files, and a changelog entry on 15 June 2026 repeats that the file is not required for visibility or rankings. Content does not need chopping into chunks, because Google’s systems understand multiple topics on one page. AI specific rewrites are pointless, since the systems handle synonyms. Chasing inauthentic mentions is warned against, and structured data is not required for generative AI search.

That last point needs care, because not required is not the same as remove it. The same guide recommends keeping structured data as part of an overall SEO strategy, and it still drives rich results and feeds the Knowledge Graph. What changes is the justification: Marking up a page so your entity is unambiguous is sound, marking it up because a vendor promised it unlocks AI Overviews is not.

One caveat: Google speaks for Google. Generative engine optimization also covers ChatGPT and Perplexity, which fetch content through their own crawlers. That llms.txt does nothing in Google Search says nothing about what other providers read.

What this changes in your quarter: The list of unnecessary tactics gets the attention, the positive half is what costs money. Google defines non commodity content by contrast, naming a first hand review built on personal experience against a summary that restates what is already available elsewhere. As a commissioning rule that reads: If a competent writer could produce the page without speaking to anyone inside your company, it is the kind of page the guide warns about. It is also the one part of this guide that a developer cannot execute for you.

Trend 3: FAQ rich results ended on 7 May 2026

Google added a deprecation notice to the FAQ rich result documentation on 8 May 2026, stating the feature would no longer appear in Google Search starting 7 May 2026, and removed the documentation entirely on 15 June 2026. On 6 January 2026 it had already removed the practice problem structured data documentation for the same reason.

The impact is smaller than the headline suggests, because Google had restricted FAQ rich results to authoritative government and health sites back in August 2023. What ended in May 2026 is the possibility of it returning, and with it the last argument for building an FAQ section purely to win space in the results.

Meanwhile rich snippets tied to commerce kept gaining properties: Merchant listing markup got Product.category and a documented sale duration on 7 July 2026 plus hasAdultConsideration on 20 May 2026, and discussion forum markup gained properties on 24 March 2026. Google is pruning informational rich results and investing in transactional and community ones.

For a JSON-LD strategy that means splitting the budget by purpose: Markup that renders a visual result earns its keep on click through rate, markup that describes your organization earns its keep on entity clarity. FAQPage markup now sits in neither bucket for Google, which is a reason to stop expanding it rather than to rip it out.

What this changes in your quarter: Move the markup hours to the types Google spent 2026 extending, because each of those additions is a concrete field somebody has to fill. Product.category accepts both Text and CategoryCode, so the category tree you already maintain for Merchant Center can be reused instead of invented. The documented sale duration gives a promotion an explicit end, which is the cheapest way to stop a finished discount from lingering in your markup. Discussion forum and QA markup gained properties that describe comment thread structure, which matters to anyone running a community. And any backlog ticket that reads add FAQPage markup to win a rich result can simply be closed.

Trend 4: Preferred sources reached AI Overviews and AI Mode

Preferred sources let a user nominate the sites they want to see more of. Google published the publisher documentation on 30 January 2026, expanded the feature to every language Google Search operates in on 30 April 2026, and extended it to AI Overviews and AI Mode on 27 May 2026. Content from a preferred source becomes more likely to appear in Top Stories and gets highlighted with a “preferred” badge, now inside AI answers too.

The mechanics are unusually direct. Users pick sites at google.com/preferences/source, and publishers point readers there with a deeplink or a button built from Google’s badge assets, available in fourteen languages. Only domains and subdomains qualify, so a blog in a subdirectory cannot be nominated on its own.

This deserves more attention than it has received, because it is the first Google surface where an existing audience relationship converts directly into AI answer visibility. Everything else on this list is mediated by ranking systems. Preferred sources is earned by asking people who already read you, so put the button at the end of long articles and in the newsletter footer, not the homepage hero.

What this changes in your quarter: The build is one link, the work is placement and expectation. The pattern is https://google.com/preferences/source?q=example.com, which means a brand running shop.example.com next to its magazine has to decide which of the two it asks readers to nominate, because a pick does not travel between them. Then say the unflattering half out loud before somebody else says the flattering one. The documentation promises a badge and a better chance of being shown to the people who chose you. It promises nothing about ranking, and nothing at all about the readers who never opted in.

Trend 5: Four ranking updates between February and June

Google confirmed four ranking updates in the first half of 2026, two of them core updates, plus a dedicated Discover update.

One number in that table should change how teams react. The core updates ran 12 and 11 days, so rankings kept moving for a week and a half after each announcement. Reading a chart on day three and rewriting pages is how sites chase noise. Wait for the completion note, then compare equally long periods after and before.

The February Discover update ran 21 days, and Google expanded its Discover documentation during the rollout. Anyone whose search results traffic looks flat while a competitor grows should check whether the gap sits in Discover. Core updates otherwise keep rewarding the profile Google has described for years, where E-E-A-T stays relevant.

How long a ranking update actually runs

The dashboard is also a two year record of how much patience the job needs. Every core update Google has completed since March 2025 took between eleven and eighteen days: 13 days in March 2025, 16 days in June 2025, 18 days in December 2025, then 12 days in March 2026 and 11 days in May 2026. Spam updates behave the other way round, mostly hours to a couple of days, with one outlier in August 2025 that ran 26 days.

The cadence is worth noting as well, because the question of when the next core update lands has no schedule behind it. Google ran core updates in March, June and December 2025, then in March and May 2026, so the gaps ranged from about two months to about six. There is no quarterly rhythm to plan around, only a completion note to react to.

The more useful pattern is where the rollouts sit in the calendar. Three of those five crossed a month boundary: June 2025 ran into July, March 2026 into April, May 2026 into June. That is the mechanical reason a monthly report so often shows a step nobody can account for. The reporting period and the rollout period simply do not line up, and the report gets blamed for the mismatch.

What this changes in your quarter: Copy the start and completion dates from the dashboard into the calendar your reporting runs on. A surprising share of core update panic starts because a monthly report happened to cut straight through the middle of one.

Trend 6: Search agents shipped, and the guidance is still thin

At I/O on 19 May 2026 Google made Gemini 3.5 Flash the global default model in AI Mode, announced information agents that watch the web continuously for updates matching a user’s criteria, and showed Search generating custom interfaces for a task. The agents launch first for premium subscribers, with booking of local services as the next step.

Google’s generative AI guide points site owners at agent friendly best practices and at emerging protocols such as the Universal Commerce Protocol. That is the extent of the concrete advice today: No agent specific markup, no agent report in Search Console, no public data on how much traffic agents send. The right posture is watchful rather than expensive. An agent needs what a crawler needs: A page it can reach, a robots.txt that does not lock it out at the edge, and facts stated plainly enough to act on. A booking flow that only works after three JavaScript interactions defeats an agent, and a share of your human visitors with it. That overlap is why agentic SEO today means doing the technical basics properly. Put agents on the watch list, not in the budget.

What this changes in your quarter: Book a review date, not a budget line. Google names two thresholds worth waiting for: Agents start with premium subscribers and booking of local services is the announced next step, and the Universal Commerce Protocol is described as emerging rather than shipped. When booking opens beyond premium, or when that protocol stops being called emerging, the topic has earned a proper look and a budget conversation.

Trend 7: Enforcement moved into AI answers and into the agency market

The quietest thread of 2026 runs through the spam documentation. On 15 May 2026 Google clarified that its spam policies apply to generative AI responses in Google Search, which closes the argument that AI answers are unpoliced. On 13 April 2026 it added a spam policy for back button hijacking, and twice in April it clarified that spam reports may lead to manual action.

On 24 July 2026 Google extended its review snippet guidelines to cover fake and undisclosed incentivized reviews. That lands on local marketing, where discount for a review campaigns are still common: Google reviews collected through undisclosed incentives now put the whole review snippet at risk. Anyone running local SEO across several locations should audit how reviews are solicited.

The most unusual entry came on 5 June 2026, when Google published guidance on evaluating third party SEO tools and services and pointed readers from it to its older page on whether you need an SEO at all. A buyer’s checklist in the same half year as a myths section about AEO and GEO reads as a deliberate pair, aimed at the gap between what vendors promise about AI visibility and what Google says it measures. Agencies selling shortcut tactics now argue against Google’s own documentation.

The page is blunter than its title suggests. It states that third party tools have no access to Google’s internal ranking data and cannot guarantee performance, that Google does not evaluate third party services so any claim of endorsement deserves suspicion, and that good advice either marks its claims as opinion drawn from data and experience or cites official Google guidance. That last sentence is a usable test for every SEO deck you are handed. It also puts vendor scores in their place, domain authority included: A useful outside estimate, never a number Google ranks by.

The four policies behind the 2026 spam updates

Google’s spam policy page, last updated on 15 May 2026, is the standing list that the March and June spam updates enforce. Four entries carry most of the weight, and three of them describe things that are still actively sold.

Site reputation abuse is the one that turns up in commercial offers, usually under the name parasite SEO, and the policy explicitly reaches content from users, freelancers and white label providers. So the question about a coupon subfolder, a sponsored section or a rented subdomain is not whether you approved the copy. It is whether the arrangement exists mainly to borrow your ranking signals. The same reading applies to any third party content living on your domain.

Scaled content abuse deserves a careful reading, because it is the policy most often misquoted as a ban on AI writing. The wording targets pages produced at volume for rankings rather than for readers, and names generative AI as one way of producing them. Volume without value is the offense; the tool is not.

Two clarifications in April 2026, on the 14th and the 23rd, tightened the loop between reports and consequences: Google stated that it may use submitted spam reports to take manual action, and set out when and why it does. A spam report is no longer only an input to the ranking systems.

What this changes in your quarter: Two lists, one afternoon. Everything a third party publishes on your domain, read against the site reputation abuse wording rather than against whether you liked the article. And every incentive your review collection touches, read against the sentence added on 24 July 2026: Reviews written in exchange for a benefit such as money, discounts, vouchers or free products need a clear and prominent disclosure, or they do not belong on the page or in the markup. The consequence is unusually specific for a Google policy. A structured data manual action makes Google ignore the markup on that page while the page itself stays in the results, so the loss is your stars, not your ranking.

What did not change, which is most of it

Seven changes in seven months sounds like upheaval, yet the work list barely moved. The eligibility rule is unchanged and still the foundation: Google’s AI features documentation, last updated on 10 December 2025, states that to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet. It adds that there are no additional requirements, no special optimizations and no need to create new machine readable files, AI text files or markup.

The quality guidance is unchanged too. Google still asks for helpful, reliable, people first content, now phrased as non commodity content with a distinctive perspective, and the technical requirements still read crawlability, clean site structure, semantic HTML and reduced duplicate content. Core Web Vitals kept the same three metrics throughout. So the accurate summary of 2026 is narrow: Google changed how visibility is measured, which rich results exist, who can nominate a source and where enforcement reaches, not what makes a page qualify.

Two things follow. Anything you were doing because Google documented it is still right. Anything you were doing because a vendor said AI search rewards it never had a documented basis to begin with. The eligibility sentence is the test that separates them: If a tactic does not make a page more likely to be indexed, more likely to be shown with a snippet, or more worth citing once it is shown, it was never about AI visibility in the first place.

The list of things people expected to change and that did not is just as instructive. There is still no separate AI ranking system to aim at, because the generative AI features run on the core Search ranking and quality systems. There is still no way to see what an AI answer said about you, only how often you were shown inside one. And there is still no lever that lifts a page into an AI answer without also lifting it in ordinary search, which is why the two disciplines have not split into separate budgets and probably will not.

The practical consequence is a budget one. If most of the work list is unchanged, most of a 2026 SEO budget belongs where it belonged in 2025: In content worth citing and in the technical work that lets a machine reach it, from indexability through to duplicate content and page speed. Everything above is worth roughly a day of reading and a handful of tickets. It is not a strategy reset, and any offer that presents it as one is selling the churn rather than the change.

Your priorities for the rest of 2026

Six moves, in the order that pays. Each one belongs to a trend above, where the detail and the source sit; this is the sequence to work them in. The first three are cheap and immediate, the last three need a decision.

1. Check the generative AI performance report. The URLs it names are your list of pages Google already trusts enough to cite.
2. Re-audit indexability and snippet eligibility. A stray noindex or nosnippet removes a page from AI answers silently. An SEO audit finds this in hours.
3. Add a preferred sources button. Fifteen minutes of work, Google supplies the badge assets, and it is the only lever where your readers push you into AI answers directly.
4. Decide about FAQPage markup deliberately. Stop building pages around it, and check what else consumes the data before deleting.
5. Delete what Google’s guide contradicts. Roadmap items like llms.txt for Google visibility, content chunking or AI specific rewrites now have a documented rebuttal. Move the time to content with a point of view.
6. Watch agents, do not budget for them. Revisit once Google ships agent reporting.

None of this needs a new tool category. It needs someone to read the changelog, decide which entries touch your site and act on the four or five that do. That is the job in our SEO services.

Frequently asked questions about the SEO trends of 2026

What are the biggest SEO trends in 2026?

AI visibility reporting in Search Console, a generative AI opt out, Google’s own guide to optimizing for generative AI, the end of FAQ rich results on 7 May 2026, preferred sources reaching AI answers, four ranking updates, and search agents shipping at Google I/O.

Is SEO still relevant in 2026?

Yes, and Google says so in writing: Its generative AI guide states that SEO best practices stay relevant because the generative AI features in Search are rooted in Google’s core ranking systems. A page still has to be indexed and eligible for a snippet to be cited in an AI answer.

Will AI replace SEO?

No. AI changes where the answer appears, not how a page qualifies, and Google’s documentation calls optimizing for generative AI search a form of SEO. What did change is measurement: Impressions inside AI answers show up in Search Console, clicks do not.

Does Google use llms.txt?

No. Google confirmed on 15 June 2026 that llms.txt files are not required for visibility or rankings in Google Search, and its generative AI guide states that Google Search itself does not use them. Other providers may read the file, so publishing one is optional rather than harmful.

Should I remove my FAQPage structured data?

You do not have to. Google stopped showing the FAQ rich result on 7 May 2026, so the markup no longer earns a visual result, but it still describes your page correctly and other systems read it. What should stop is building pages around it.

How do I track visibility in AI Overviews and AI Mode?

Open the generative AI performance report in Search Console. It shows impressions of your URLs in AI Overviews and AI Mode, split by page, country, device and date. The rollout is still partial.

How long does a Google core update take?

Between eleven and eighteen days for every core update Google has completed since March 2025, according to the Search Status Dashboard. The March 2026 core update ran 12 days, the May 2026 one 11 days. Spam updates are far shorter, usually hours to a couple of days. Wait for the completion note before you judge the effect.

Does Google penalize AI generated content?

Not as such. Google’s spam policy targets scaled content abuse, defined as many pages generated primarily to manipulate rankings rather than to help users, and it names generative AI as one way of producing them. Volume without value is the violation, not the tool that produced the text.

What is site reputation abuse?

Publishing third-party content on a host site mainly because of that host’s already established ranking signals, so it ranks better than it could on its own. Google’s spam policy reaches content from users, freelancers and white label providers alike. The tactic is usually sold under the name parasite SEO.

Sources

• Google Search Central: “AI features and your website“, last updated 10 December 2025.
• Google Search Central: “Optimizing your website for generative AI features on Google Search“, last updated 10 July 2026.
• Search Central Blog: “A new resource for optimizing for generative AI in Google Search“, 15 May 2026.
• Search Central Blog: “Introducing Search Generative AI performance reports in Search Console“, June 2026.
• Google: “New opportunities, control and insights for website owners“, 3 June 2026.
• Google: “Google Search’s I/O 2026 updates: AI agents and more“, 19 May 2026.
• Google Search Central: “Latest Google Search documentation updates“, entries from January to July 2026.
• Search Console Help: “Generative AI performance report (Search)“, 2026.
• Google Search Central: “Preferred sources“, 2026.
• Google Search Status Dashboard: “Ranking updates history“, accessed July 2026.
• Google Search Central: “Creating helpful, reliable, people-first content“, 2026.
• Search Central Blog: “Changes to HowTo and FAQ rich results“, 8 August 2023.
• Google Search Central: “Spam policies for Google web search“, last updated 15 May 2026.
• Search Central Blog: “Introducing a new spam policy for back button hijacking“, April 2026.
• Google Search Central: “Google Search’s guidance on third-party SEO tools and advice“, last updated 5 June 2026.
• Google Search Central: “Review snippet structured data“, last updated 24 July 2026.