# AI SEO in 2026: What Changed, What Didn't, and What to Do Now

- URL: https://agricidaniel.com/blog/ai-seo-in-2026
- Author: [Agrici Daniel](https://agricidaniel.com/about)
- Published: 2026-11-18
- Updated: 2026-11-18
- Category: AI & SEO

AI SEO in 2026 is still SEO, with fewer clicks per search and new places to be cited. What Google published, what four research groups measured, and a 10-step checklist.

**AI SEO in 2026 is still SEO, with fewer clicks per search and new places to be cited.** Google says in writing that AI Overviews and AI Mode need no special optimization, no llms.txt and no new schema. At the same time, the two most recent click studies I checked measured the top organic result losing about half its clicks (52% and 58%) when an AI Overview sits above it. Both facts are true at once, and the work in 2026 is living with both.

This guide puts the sources side by side: what Google actually published this year, what four research groups measured and how, what you can now see in Search Console, and the short list of things I would do on a site today. I build open-source SEO tools, including claude-seo, so where I mention my own tools, read that as a maker talking, not an independent review.

-   \-58% top result CTR with an AI Overview (Ahrefs, Dec 2025 data)
-   \-52% position 1 CTR, US desktop (Advanced Web Ranking, July 2026)
-   8% vs 15% visits with a result click, with and without an AI summary (Pew)
-   +120% organic clicks per impression when your page is cited in the AI Overview (Seer)

_Four independent measurements with different methods and periods, so they are not directly comparable. Sources and methods are in the table further down._

In this guide

-   Key takeaways
-   What is AI SEO in 2026?
-   What Google said this year
-   What happened to clicks
-   Why being cited matters
-   How a page ends up in an AI answer
-   How to measure AI search in 2026
-   llms.txt, WebMCP and AI agents
-   What I run on this site
-   The AI SEO checklist for 2026
-   Tools that help
-   FAQ

## Key takeaways

-   Google's official position, published in May 2026: optimizing for its generative AI features "is optimizing for the search experience, and thus still SEO."
-   Clicks dropped where AI Overviews appear. The measured drop for the top result runs from 34.5% to 58% depending on the study and the year.
-   Being cited inside the AI Overview recovers part of that loss. Seer measured 120% more organic clicks per impression for cited pages than for pages that were not cited.
-   Search Console now has a generative AI report for every site, but it shows impressions, not clicks.
-   llms.txt neither helps nor harms you in Google Search. Build it for other systems if you want, not for Google rankings.

## What is AI SEO in 2026?

AI SEO is the work of making your pages easy to find, understand and cite when a search engine or an AI assistant writes an answer instead of only listing links. In practice that means Google AI Overviews and AI Mode, plus assistants like ChatGPT, Claude, Perplexity and Gemini that search the web before they answer.

You will also see it called GEO (generative engine optimization) or AEO (answer engine optimization). Google's announcement of its new guide lists a section on mythbusting AEO and GEO misconceptions ([Search Central Blog](https://developers.google.com/search/blog/2026/05/a-new-resource-for-optimizing)), and the guide's conclusion is blunt: from Google Search's point of view, this is still SEO ([Google Search Central](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide)). The labels matter less than the job: be the page an answer engine trusts enough to quote, and give the reader a reason to click through.

## What Google said this year

On May 15, 2026, John Mueller announced a new Google guide, "Optimizing your website for generative AI features on Google Search" ([Search Central Blog](https://developers.google.com/search/blog/2026/05/a-new-resource-for-optimizing)). Read it together with Google's page on AI features and your website. Here is what the two pages say, in Google's words:

-   **No special optimization.** "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." ([AI features and your website](https://developers.google.com/search/docs/appearance/ai-features))
-   **No new files or schema.** "You don't need to create new machine readable files, AI text files, or markup to appear in these features. There's also no special schema.org structured data that you need to add." (same page)
-   **No chunking.** "There's no requirement to break your content into tiny pieces for AI to better understand it." ([AI optimization guide](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide))
-   **Structured data has to match the page.** Google lists "making sure your structured data matches the visible text on the page" among the fundamentals that still apply.
-   **Original content wins.** The guide asks for "valuable, non-commodity content" and sums it up as one principle: "focus on what your visitors would enjoy, find helpful, and feel satisfied with after visiting your website."

My reading: Google is telling you that the shortcut industry around AI search does not apply to Google. If someone sells you an "AI schema pack" or a special file for AI Overviews, ask them to show you where Google says it is used. Google's documentation says the opposite.

What Google did not say is just as useful. It does not promise that good SEO brings back the clicks. That part is in the data.

## What happened to clicks

Every serious measurement I found points the same way: when an AI Overview appears, fewer people click the organic results. The size of the drop depends on the method, the period and the query mix.

Position 1

Organic click-through rate for the first result, Google US desktop, July 2026

| Search results page | Position 1 CTR |
| --- | --- |
| All results pages (baseline) | 20.02% |
| Results pages with an AI Overview | 9.69% |

_Source: [Advanced Web Ranking organic CTR study](https://www.advancedwebranking.com/seo/organic-ctr), July 2026 data, more than 1.75 million US results pages a month. AI Overviews appeared on 43% of the sampled desktop pages. A vendor estimate, not a peer-reviewed study._

Five AI Overview click studies, their methods and results

| Study | What it measured | Result |
| --- | --- | --- |
| [Pew Research Center](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/), July 2025 | Real browsing of 900 US adults in March 2025, 68,879 Google searches | A result link was clicked in 8% of visits with an AI summary and 15% without. Links inside the summary got clicked in 1% of visits. |
| [Ahrefs](https://ahrefs.com/blog/ai-overviews-reduce-clicks/), April 2025 | 300,000 keywords, desktop CTR of the top result, March 2024 vs March 2025 | 34.5% lower CTR with an AI Overview |
| [Ahrefs update](https://ahrefs.com/blog/ai-overviews-reduce-clicks-update/), February 2026 | Same design, December 2023 vs December 2025 | 58% lower CTR for the top result with an AI Overview |
| [Advanced Web Ranking](https://www.advancedwebranking.com/seo/organic-ctr), 2026 | US results pages, July 2026, position 1 on desktop | 20.02% CTR overall, 9.69% with an AI Overview, a 52% drop |
| [Seer Interactive](https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-2026-update), April 2026 | 53 brands, 5.47 million queries, January 2025 to February 2026 | Organic CTR with an AI Overview rose from 1.31% (December 2025) to 2.36% (February 2026), still below 3.82% without one |

Two things stand out. First, the Ahrefs re-run found a bigger drop than its first study, which suggests the effect has not faded as people got used to AI Overviews, although the two studies cover different periods. Second, Seer's numbers moved up between December and February, which tells you this is not a straight line and monthly checks beat yearly assumptions.

What this means for you: a page can hold position 1 and still lose half its clicks. If your report says "rankings flat, clicks down," check whether those queries started showing an AI Overview before you blame the content.

## Why being cited matters

The same Seer study separated pages that the AI Overview cited from pages it did not. Being cited delivered 120% more organic clicks per impression than not being cited, and through 2025 a cited page earned 2 to 5 times the organic CTR of an uncited one ([Seer Interactive](https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-2026-update)). Even cited pages still trailed queries with no AI Overview by 38% on informational searches.

So the goal shifts a little. You still want the ranking, but on a query with an AI Overview, the citation is what separates a page that keeps some clicks from a page that keeps far fewer. This is the most practical definition of AI SEO I can give you: earn the citation, then earn the click.

## How a page ends up in an AI answer

Google explains one mechanism by name. Both AI Overviews and AI Mode "may use a 'query fan-out' technique," which Google describes as issuing multiple related searches across subtopics and data sources to build the response ([Google](https://developers.google.com/search/docs/appearance/ai-features)). In other words, one question from the user becomes several searches behind the scenes, and the answer is built from what those searches return.

Query fan-out

From one question to a cited answer

1.  1
    
    **Your page is crawlable and indexed**
    
    Google requires a supporting link to be indexed and eligible to show with a snippet. It also recommends allowing crawling in robots.txt and by your CDN, and linking the page internally.
    
2.  2
    
    **A user asks one question**
    
    Often longer and more specific than a classic keyword.
    
3.  3
    
    **The model fans out**
    
    It runs several related searches across subtopics of that question.
    
4.  4
    
    **Results come back**
    
    Pages that answer one of those subtopics clearly are candidates.
    
5.  5
    
    **The answer cites sources**
    
    A cited page gets a link inside the answer, and with it a chance at the click.
    

_Steps 1 and 3 are from Google's documentation. How Google picks which result to cite is not published; steps 2, 4 and 5 are my simplified reading, not Google's description._

The practical consequence is my inference, not a Google statement: a page that covers the obvious follow-up questions of a topic has more ways to match one of the fan-out searches than a page that answers one narrow question. That is a case for depth and clear sections, not for splitting one topic into dozens of thin pages.

## How to measure AI search in 2026

For the first time you can see Google's AI features in your own data. On June 3, 2026, Google announced generative AI performance reports in Search Console, and on August 31, 2026 it rolled them out to all websites worldwide ([Search Central Blog](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports)).

-   **What it shows:** impressions in generative AI features in Search and Discover, broken down by page, country, device and date.
-   **What it does not show:** Google's announcement lists no clicks or CTR for these features. Clicks from AI Overviews and AI Mode are counted in your normal Performance report under the Web search type, mixed with everything else ([Google](https://developers.google.com/search/docs/appearance/ai-features)).
-   **What it never shows:** ChatGPT, Claude, Perplexity or any other assistant. Those visits only appear as referrals in your own analytics, when the assistant passes a referrer at all.

A simple routine that works with what exists: once a month, export the generative AI report next to the normal Performance report for the same pages. Pages with many AI impressions and falling clicks are your first candidates for a citation check and a rewrite.

## llms.txt, WebMCP and AI agents

**llms.txt** is a plain Markdown file that summarizes a site for AI systems. Google's answer is precise: "Doing so will neither harm nor help your site's visibility or rankings in Google Search, as Google Search ignores them." The same guide calls it "completely fine" to keep one for other services that use it ([Google](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide)). So build it if it helps agents or tools you care about, and never sell it to a client as a Google ranking fix.

**WebMCP** is the newer idea. Chrome describes it as "a proposed web standard to help you build and expose structured tools for AI agents," so an agent can call a site's search or form directly instead of guessing at buttons. It is an origin trial from Chrome 149, and Chrome says it is "under active discussion and subject to change in the future" ([Chrome for Developers](https://developer.chrome.com/docs/ai/webmcp)). Treat it as an experiment worth watching, not as a ranking factor. Nothing I found says it affects search visibility.

## What I run on this site

Here is the setup on agricidaniel.com, which you can check yourself because every part is public. I am showing it as a worked example, not as proof that it brings traffic. I have no measured result from these files on their own.

-   **robots.txt with content signals.** Search crawlers and named AI crawlers such as GPTBot, ClaudeBot and PerplexityBot are allowed, and so is Google-Extended, which Google describes as a product token, not a separate crawler, that controls use of content for Gemini training and grounding, all with `Content-Signal: search=yes, ai-input=yes, ai-train=yes` repeated in each group of `agricidaniel.com/robots.txt`. That is my choice for a site whose job is to be found and quoted. A publisher that sells its content may choose the opposite.
-   **llms.txt.** A short site summary with dated facts at `agricidaniel.com/llms.txt`. Per Google, it does nothing for Google Search; it is there for other agents.
-   **A Markdown copy of every post**, linked from the page head, so an agent can read the article without the layout.
-   **WebMCP tools** in the origin trial, so a browser agent can search the blog and list the open-source projects.
-   **Structured data that matches the page:** Article, FAQPage and BreadcrumbList, built from the same text you read.

The boring parts carry the weight: one canonical URL per page, a clean sitemap, headings that say what the section answers, and sources linked next to the claims. Those are the same fundamentals Google lists, and none of them is new.

## The AI SEO checklist for 2026

1.  **Check access first.** Make sure robots.txt and your CDN or firewall do not block Googlebot or the AI crawlers you want. Then decide, on purpose, which AI crawlers you allow.
2.  **Keep pages indexable and snippet-eligible.** Google's controls for AI features are the same as for normal results: `nosnippet`, `data-nosnippet`, `max-snippet` and `noindex`. Use them only where you mean to limit what Google shows.
3.  **Answer first, then explain.** Start each section with the direct answer to its heading, then give the detail. It helps a reader in a hurry and gives an answer engine a clean passage to quote.
4.  **Add something only you have.** Your own numbers, screenshots, tests or experience. Google asks for "non-commodity content," and a page that repeats ten other pages gives an AI answer no reason to cite it.
5.  **Cover the follow-up questions.** Map the questions a reader asks next and answer them on the same page or link to the page that does.
6.  **Cite your sources next to the claim.** Dated, linked, with the method when it matters. It helps the reader check you, which is the point.
7.  **Keep structured data honest.** Mark up what is visible on the page and nothing else. No special AI schema exists.
8.  **Measure monthly.** Search Console's generative AI report for impressions, the Performance report for clicks, your analytics for assistant referrals.
9.  **Check citations on the queries that matter.** Search them yourself, or use a tool, and note whether your page is cited in the AI Overview. That is the number Seer's data says moves clicks.
10.  **Skip the shortcuts.** No llms.txt-for-rankings, no splitting pages into fragments, no AI schema packs. Google says none of them is used.

## Tools that help

You can do most of this checklist with free tools. These are the ones I build and use, so weigh my opinion accordingly:

-   [claude-seo](https://github.com/AgriciDaniel/claude-seo) is a free, MIT-licensed SEO toolkit for Claude Code with 26 skills and 19 specialist agents (version 2.4.0, counts checked 2026-11-11). It includes a GEO check for crawler access and citation readiness. The full walkthrough is in [the Claude Code SEO stack](https://agricidaniel.com/blog/claude-code-seo-stack).
-   If you work in ChatGPT or Codex instead, there is [ChatGPT SEO](https://agricidaniel.com/blog/chatgpt-seo-plugin) and [codex-seo for OpenAI Codex CLI](https://agricidaniel.com/blog/codex-seo-openai-codex-cli).
-   [DataForSEO Agent](https://agricidaniel.com/blog/dataforseo-agent) can check which of your keywords show an AI Overview and whether it cites your site. The app is free; the DataForSEO data is paid per call.
-   For a broader free toolkit, see [free SEO audit tools that actually work](https://agricidaniel.com/blog/free-seo-audit-tools), and for a worked example of an AI-run audit, [the Jev SEO audit](https://agricidaniel.com/blog/jev-seo-ai-seo-audit).

## Frequently asked questions

### Is AI SEO different from regular SEO?

For Google, no. Google's 2026 guide says optimizing for its generative AI features is still SEO, and its documentation says no special optimization is needed for AI Overviews or AI Mode. What changed is the result: fewer clicks per search where an AI Overview appears, and a new goal of being cited inside the answer.

### Do I need an llms.txt file for Google?

No. Google says its Search ignores llms.txt and that the file will neither help nor harm your visibility in Google Search. You can still publish one for other AI tools and agents that read it.

### How much traffic do AI Overviews take?

It depends on the query and the study. Ahrefs measured 58% lower CTR for the top result with an AI Overview in its February 2026 update, Advanced Web Ranking measured a 52% drop for position 1 on US desktop in July 2026, and Pew found result clicks in 8% of visits with an AI summary versus 15% without.

### Can I see AI Overview traffic in Search Console?

Partly. Since August 31, 2026 every site has a generative AI report with impressions by page, country, device and date. Clicks from AI features are counted inside the normal Performance report under the Web search type and are not shown separately.

### Is there special schema for AI Overviews?

No. Google says there is no special schema.org structured data to add for AI features. Keep using normal structured data and make sure it matches the visible text on the page.

## Related reading

-   [The Claude Code SEO stack](https://agricidaniel.com/blog/claude-code-seo-stack)
-   [DataForSEO Agent: check AI Overviews and rankings from the terminal](https://agricidaniel.com/blog/dataforseo-agent)
-   [Free SEO audit tools that actually work](https://agricidaniel.com/blog/free-seo-audit-tools)
-   [Local SEO Brain for small businesses](https://agricidaniel.com/blog/claude-code-local-seo-brain)

## Start with one page

Pick the page that brings you the most search traffic. Search its main query, see whether an AI Overview appears and whether it cites you, and open the generative AI report for that page in Search Console. That one check tells you more about your AI SEO in 2026 than any framework.

Want to work through it with other builders? Join the [free AI Marketing Hub](https://skool.com/ai-marketing-hub), or [AI Marketing Hub Pro](https://skool.com/ai-marketing-hub-pro).
