SEO trends 2026: what matters in search now

Search engine optimisation is changing faster in 2026 than in previous years. The foundation stays the same: in its AI features, Google only shows pages that are indexed and eligible for a snippet. What is new is where and how your content appears, and how you measure it.

• Answers in search: since May 2026, AI Overviews and AI Mode have merged into one AI Search experience. Clicks are becoming rarer, and being cited as a source matters more.
• New metrics: Search Console, GA4 and Bing Webmaster Tools now report visibility in AI answers and visits from AI assistants separately.
• Visibility beyond Google: ChatGPT, Perplexity, Copilot and social and video platforms are becoming routes to your business in their own right.
• Quality and trust: three core updates in twelve months reward original content based on experience. Spam policies target tricks and mass-produced AI content.
• Tidier search results: Google has switched off rarely used features and several types of structured data.
• Control over AI use: since August 2026, you can opt out of AI Overviews and AI Mode in Search Console. A shared robots.txt signal for AI training is on its way.
• Technical SEO as the foundation: indexing, status codes and images still decide whether content is eligible at all.

Many search results now open with an AI Overview. For follow-up questions there is AI Mode, a conversational form of search. Since May 2026, according to Google, the two have merged into one continuous AI Search experience, live worldwide on desktop and mobile. Google says AI Mode now has more than a billion monthly users. Both features summarise answers and link to the sources they draw on. According to Google, people ask questions in AI Mode that are almost three times as long as in classic search.

To build an answer, the system runs several related searches at the same time, a technique Google’s guide calls query fan-out. Pages that answer one of these sub-questions well can therefore be cited as sources too.

For websites, this means fewer clicks on questions that a summary already answers. Studies differ on how large the drop is. Using data from December 2025, Ahrefs found in a modelled comparison that the top organic result had a click-through rate around 58% lower when an AI Overview appeared. Seer Interactive reports a 61% fall in organic click-through rate for informational queries with an AI Overview between June 2024 and September 2025. The same study also shows the other side: brands cited in an AI Overview received around 35% more organic clicks than brands that were not. Seer points out that this is a correlation and does not prove the cause. Both studies use their own data sets and methods. They show the direction, but their figures cannot be applied to your own website.

In August 2025, Google wrote that AI in Search was sending more quality clicks to websites, meaning clicks after which users do not quickly click back. Total organic click volume, it said, had been relatively stable year on year. Google gives no figures to back this up.

What you should do:

• Answer your audience’s questions fully and with your own knowledge: your own figures, examples and experience. Google names unique, non-commodity content as the most important lever.
• For your ten most important topics, collect the follow-up questions that customers ask your sales and support teams, and check whether your page answers each of them.
• Once a month, check for your 20 most important queries whether an AI Overview appears and whether your page is linked there as a source. Record the result in a spreadsheet so that you can see changes over time.
• Judge content by enquiries, sales and sign-ups. In GA4, the landing page report shows these alongside your key events.

For a long time, there was no way to tell whether a visit came from an AI answer or from a normal result. There are now dedicated tools for this:

• Generative AI performance reports in Search Console: since June 2026, they show how often URLs from your website appeared in AI Overviews, AI Mode and AI features in Discover, broken down by page, country, date and, for Search, device. According to Google, they have been available to all websites since 31 August 2026. The report shows impressions; it does not show clicks. More on this in our post on the Search Console generative AI report.
• “AI Assistant” channel in GA4: Google Analytics 4 groups visits from ChatGPT, Gemini, Copilot and other assistants in a default channel of their own. According to Google’s help centre, clicks from AI Overviews and AI Mode still count as organic search.
• Branded queries filter: since November 2025, Search Console has been introducing a filter that separates queries containing your brand name from all others. Since March 2026 it has been available to all eligible sites, but only for top-level properties, not for subfolders or subdomains, and only where there are enough queries. This shows whether your brand awareness is growing even when clicks on general questions fall.
• Custom annotations: since November 2025, you can add your own annotations to Search Console charts, for example for core updates, a relaunch or new content. Everyone with access to the property can see them.
• Multimodal search: since September 2026, Search Console has reported searches with Google Lens, Circle to Search, uploaded images and “Search this image” in Chrome via a filter of its own.
• AI report in Bing Webmaster Tools: since February 2026, Microsoft has shown in Bing Webmaster Tools, as a public preview, how often your pages are cited as sources in Copilot and in Bing’s AI summaries, and which queries the AI used to find them. This report does not include clicks either.

On rheinwunder.com, around 0.7% of the sessions measured in GA4 between June and October 2026 came from AI assistants, while organic search brought in just under half. GA4 only measures visits with consent on our site, so we only look at shares. On our site, clicks from AI Overviews can no longer be identified via text fragments in the URL either: in our sample of 5 November 2026, none of the three links we tested arrived with a text fragment. The details are in our post Tracking AI Overviews in GA4.

What you should do:

• Create a report for the “AI Assistant” channel in GA4 and track enquiries and sales there.
• Compare a page’s AI impressions with its clicks in the performance report. Pages with many AI impressions and few clicks are candidates for revision.
• Track branded queries as a metric of their own.
• Add core updates, relaunches and major content changes as annotations in Search Console. This makes it quicker to explain fluctuations later.
• Verify your website in Bing Webmaster Tools. You can import sites from Search Console.

People no longer search on Google alone. ChatGPT, Perplexity, Copilot and Gemini answer questions with sources of their own, and many people search directly on Instagram, TikTok and YouTube. For visibility in AI answers, the term generative engine optimisation (GEO) has become established.

Since 24 August 2026, ChatGPT has also shown ads in Germany and many other European countries, initially in the free version and the Go plan. Alongside unpaid citations in AI answers, there is now a paid route into ChatGPT. Our page on ChatGPT Ads explains how ads work there.

Google is matter-of-fact about this. In its guide to optimising for generative AI from May 2026, Google writes that optimising for generative AI search is still SEO. Other AI assistants read websites through their own crawlers and fetchers, which you can allow or block in robots.txt. Our article on AI crawlers explains what they are called and what they fetch.

For social and video content, Google introduced platform properties in Search Console in July 2026 and has been rolling them out gradually. They show how your posts on Instagram, TikTok, X and YouTube perform in Google Search, Discover and News.

What you should do:

• Check regularly with a fixed set of questions whether and how AI assistants mention your brand. Our AI visibility tool does this, for example.
• Allow the crawlers of AI search services if you want to appear there as a source, for example OAI-SearchBot for ChatGPT search and PerplexityBot for Perplexity.
• Connect your social and video profiles to Search Console as platform properties.

Google rolled out three core updates within twelve months: in December 2025, March 2026 and May 2026, plus several spam updates, most recently in September 2026, as the Search Status Dashboard shows. February 2026 brought a separate core update for Discover, initially for English-speaking users in the US. According to Google, it is meant to reduce sensational content and clickbait and to show in-depth, original content from sites with expertise. The direction has been the same for years: Google rewards content based on first-hand experience and recognisable expertise, summed up as E-E-A-T.

Many companies now write content with AI. Google allows this as long as the content meets the same standards as everything else. According to Google, using AI to generate many pages without adding value violates its spam policy on scaled content abuse. In its guidance on AI-generated content, Google asks you to fact-check all AI content by hand before publishing, including page titles, meta descriptions, structured data and alt text. Online shops must label AI-generated product images and product descriptions as AI-generated in Merchant Center.

Two rules are new or changed in 2026:

• Manipulated back button: since 15 June 2026, Google has treated it as spam when a page manipulates the browser’s back button (“back button hijacking”). This can also be caused by included libraries or advertising networks.
• Site reputation abuse: since 30 August 2026, manual actions against third-party content on trusted websites no longer apply for users in the European Economic Area. Outside it, they still affect the section concerned. For this change, Google points to discussions with the European Commission and to the Digital Markets Act. The policy itself still applies, so the risk remains for businesses with audiences outside Europe.

What you should do:

• Show who is writing and where the knowledge comes from: authors with profiles, your own data, traceable sources.
• Decide who reviews AI-generated content before publication, and keep a record of the review. Do not publish pages that merely repackage what is already known.
• Revise or merge thin and duplicate pages so that several pages do not compete for the same queries.
• Check third-party scripts for interference with browser history.

Google is tidying up the results page. In June 2025, Google announced that it would gradually switch off seven types of structured data in Search, including Course Info, Claim Review, Estimated Salary and Vehicle Listing. According to Google, this did not affect rankings. In November 2025, Google announced that it would gradually remove further rarely used features.

According to Google’s guide, AI features do not need any special markup: structured data is not a requirement for generative AI search. It remains useful all the same, for example for products, reviews, organisations and breadcrumbs, as long as it matches the visible content.

What you should do:

• There is no rush to remove markup for switched-off types. It does no harm, but it no longer does anything in Google Search.
• Maintain the types Google still uses and check them in the Rich Results Test. More in our article on structured data.

Since 31 August 2026, every website can decide in Search Console whether its content appears in the AI features of Google Search. According to Google, the setting covers AI Overviews, AI Mode and AI Overviews in Discover. Sites that opt out no longer receive impressions or visits from these features. Google says it does not use the setting as a ranking signal for search results outside these AI features.

Training AI models is a separate matter. Whether Google may use your content to train its Gemini models is controlled with the Google-Extended entry in robots.txt. This entry has no effect on Google Search. According to Google, however, websites that explicitly allow Google-Extended can appear in Gemini app responses through grounding and receive clicks from there. In October 2026, Google announced support for a shared signal for AI training being developed by the IETF’s AIPREF working group. Google plans to treat an opt-out via this signal as if Google-Extended were disallowed, provided Google-Extended is not explicitly specified. Google aims to support the signal together with other AI providers in 2027. Until then, it is an announcement.

For individual pages, the familiar snippet rules still apply: nosnippet, data-nosnippet, max-snippet and noindex. The tighter you set them, the less of your content Google shows in the results and in AI answers.

What you should do:

• Leave the AI setting in Search Console at its default for as long as you want to be cited in AI answers. Before opting out, check the generative AI report to see how many impressions you would lose.
• Decide on training and search separately: Google-Extended for training, the Search Console setting for AI Overviews and AI Mode, and robots.txt for other providers’ crawlers. Many providers use separate crawlers for training, search and fetches on behalf of individual users. Our article on robots.txt shows examples.
• Only set snippet rules on pages whose content you want to hold back.

For Google to show a page in AI features, it must be indexed and eligible for a snippet. According to Google, the same technical requirements as for normal search are sufficient: Googlebot is not blocked, the page returns status code 200, and the content is indexable. Errors here therefore cost you twice: in the results and in the answers.

Images, videos and business data are gaining weight. For multimodal searches, Google recommends high-quality images and videos. In its guide to generative AI search, Google also advises local businesses and online shops to keep their Business Profile and Merchant Center details complete and up to date. Page speed remains part of page experience, which for Google also includes the Core Web Vitals. More in our article on the Core Web Vitals.

What you should do:

• Check indexing, redirects and status codes regularly, especially after a relaunch. Our page on technical SEO describes how we approach this.
• Give images meaningful file names and alt texts, and keep product data in Merchant Center and your Business Profile details up to date.

After AI that answers comes AI that acts. Google’s guide describes AI agents as systems that carry out tasks on people’s behalf, reading websites through the page structure and the accessibility tree. At I/O in May 2026, Google announced Search agents, rolling out first to Google AI Pro and Ultra subscribers. For shopping, Google introduced the Universal Commerce Protocol in January 2026, an open standard that lets people complete purchases directly in AI Mode and the Gemini app. Checkout runs on product listings from Merchant Center and is starting with retailers in the US. Google has announced a global rollout; as far as we know, there is no date for Germany yet.

Google itself advises looking into agents only once they matter for your business.

What you should do:

• Keep product data in Merchant Center complete and up to date, including price, availability and shipping costs.
• Build forms, buttons and filters so that they work with a keyboard and a screen reader. According to Google, agents read the accessibility tree, among other things, so an accessible page is easier for them to use. It also helps you meet accessibility requirements such as the European Accessibility Act.

• llms.txt: Google writes in its guide that you do not need new machine-readable files, AI text files or Markdown versions to appear in Google Search. In our own server logs, no major AI crawler fetched the llms.txt in October 2026; we do not know whether other websites see different figures. The details are in our post Who requests llms.txt?
• Breaking content into tiny chunks: according to Google, you do not need to split texts into tiny sections for AI to understand them. Well-structured texts with clear answers still help your readers.
• Keyword density: a particular frequency of a search term is not a goal. Overdoing it risks keyword stuffing, which violates Google’s spam policies.
• Tools that guarantee rankings: Google explicitly warns against third parties that promise ranking success or claim to use internal Google metrics.

In the first four weeks

• Set up measurement: open the generative AI report in Search Console, create a report for the “AI Assistant” channel in GA4, use the branded queries filter in the performance report, and verify your site in Bing Webmaster Tools.
• Record your decisions: decide deliberately on the AI setting in Search Console, Google-Extended and AI crawlers in robots.txt, and document them.
• Check the technical basics: indexing, status codes, snippet rules and third-party scripts.

Over the next three months

• Select your most important pages: the 20 pages with the most impressions and enquiries.
• Review questions and answers: which questions and follow-up questions does each page answer, and where is your own knowledge missing?
• Safeguard AI content: set up a review step for AI-generated content before new pages go live.

On an ongoing basis

• Monitor visibility in AI answers: fixed questions, fixed intervals, once a month.
• Measure the result: by enquiries and revenue per landing page. Clicks alone are not enough as a yardstick.

Prefer not to tackle this on your own? We support you with search engine optimisation and generative engine optimisation.

Changes to this article

• November 2026: fully revised for 2026, now including measurement in Search Console, GA4 and Bing Webmaster Tools, the AI setting in Search Console, the signal for AI training, mass-produced AI content, the new spam rules and an outlook on AI agents. Previously “5 important SEO trends in 2024”.