Guide · 10 min read

Google AI Mode SEO: how to show up in AI Mode

For indie developers and small SaaS teams. What Google says about AI Mode, how follow-up questions change what gets cited, and how to tell whether it's working when your traffic is small.

Updated · By the Serpkin team

In short

  • To appear in AI Mode, a page must be indexed, eligible for a snippet and not excluded with Search Console's Search generative AI setting. Google says no special markup, schema or AI files are needed.
  • AI Mode is a conversation: each question and follow-up fans out into related searches, so a focused page that answers sub-questions clearly can be cited without ranking first for the head term.
  • AI Mode and AI Overviews often cite different pages (13.7% URL overlap in an Ahrefs study), so track them separately.
  • Answer first under every heading, cover the follow-up questions people actually ask, and publish first-hand data. Google says this is still SEO.
  • Search Console's Generative AI report shows impressions only (by page, country, device, date). Clicks stay in the Web totals, and each AI Mode follow-up counts as a new query.

To show up in Google AI Mode, a page has to be indexed and eligible to show a snippet in normal Search, and your Search Console property must not be excluded with the Search generative AI control. Google says there are no other technical requirements: no special markup, schema or AI text files. So Google AI Mode SEO is mostly regular SEO, with one twist that matters for small sites. AI Mode is a conversation: one question fans out into many related searches, and people ask follow-ups. A focused page that answers the sub-questions in your niche can be cited even when it doesn't rank first for the head term.

What AI Mode is, and how it differs from AI Overviews

AI Mode is a separate, conversational search view in Google. Google introduced it in March 2025 as an experiment that "expands what AI Overviews can do" with more reasoning, built for harder questions that need exploring or comparing. You ask, get a long answer with links, then ask follow-ups to dig deeper.

AI Overviews are the short summaries that appear on top of some normal results pages. If you want the general playbook for those (myths, opt-outs, ChatGPT and Copilot), read how to get cited in Google AI Overviews. This page covers what is different about AI Mode.

AI Overviews vs AI Mode (Google's documentation, plus Ahrefs' December 2025 study where marked)
AI OverviewsAI Mode
Where it appearsOn top of some normal results pagesA separate conversational view
Shown whenGoogle judges it adds to the classic resultsThe user switches to AI Mode
Follow-upsNo, it is one summaryYes, the conversation continues
Query fan-outCan use itCan use it, on every follow-up too
Answer length (Ahrefs)ShorterAbout 4x longer, with more named entities
Search Console positionOne position shared by all its linksSame method as a normal results page

Google says the two features "may use different models and techniques", so the links can differ. How much? Ahrefs compared answers to the same US queries (September 2025 data, about 540,000 query pairs) and found the cited URLs overlapped only 13.7% of the time, even though the answers said similar things (86% semantic similarity on average). It is one vendor's study, but the takeaway is safe: being cited in an AI Overview doesn't mean you are cited in AI Mode for the same question, and the reverse.

Query fan-out and follow-ups

Google describes query fan-out as issuing several related searches at once across subtopics and data sources, then combining the results into one response. In AI Mode, each follow-up starts the process again with the conversation as context. That is why a page can be cited for a question it never targets by name.

  1. 1First question“Best way to send transactional email from a Next.js app?”
  2. 2Fan-outRelated searches: providers, pricing, deliverability, setup, limits.
  3. 3Follow-up“Which one has a free tier for 3,000 emails a month?”
  4. 4New fan-outPricing pages and comparisons with current numbers get pulled in.
How one AI Mode conversation reaches many pages (illustrative example). Each turn fans out into related searches, and any well-answered section can be among the links.

For a small site, this is the opening. You may never outrank the big review sites for the first question, but the follow-up ("free tier for 3,000 emails", "does it work with Server Actions") is often a narrow, specific question that few pages answer clearly.

What makes a page eligible

Google's AI features and your website page lists the requirements. There are three, and you control all of them:

  • Indexed. The page is in Google's index. Check with URL Inspection in Search Console.
  • Snippet-eligible. No nosnippet, no max-snippet:0, and no noindex in the robots meta tag or the X-Robots-Tag header. Googlebot is allowed in robots.txt.
  • Not opted out. Search Console has a property-level setting under Settings > Search generative AI. Set to exclude, it removes the property's content from AI Overviews, AI Mode and Discover's generative AI features, usually within one to two days. Google says it isn't used as a ranking signal for the rest of Search, and it doesn't affect AI training (that is Google-Extended). Child properties can inherit or override the parent's setting.

Google also says you don't need new machine-readable files, AI text files or special schema to appear. (Some AI tools read llms.txt; Google Search doesn't. More in do you need llms.txt?) Eligibility isn't a promise: Google says indexing and serving are never guaranteed.

Terminal
# Robots header and meta on the live page (not your source code)
curl -sI https://your-site.com/pricing | grep -i x-robots-tag
curl -s https://your-site.com/pricing | grep -io '<meta[^>]*robots[^>]*>'

# Is the main answer in the server HTML?
curl -s https://your-site.com/pricing | grep -c "per month"

What AI Mode means for clicks

Expect fewer clicks per appearance than a classic blue link, and plan for it. The best public numbers are about AI Overviews, not AI Mode, so treat them as a direction rather than a forecast:

  • Ahrefs (February 2026, December 2025 data): when an AI Overview was present, the top-ranking page's average click-through rate was about 58% lower than expected. Desktop data only, and correlation, not proof.
  • Pew Research Center (July 2025, March 2025 browsing data from 900 US adults): people clicked a traditional result in 8% of visits when an AI summary appeared, versus 15% when none did. They clicked a link inside the summary in about 1% of visits.

AI Mode answers are longer, so the reader often gets what they need without leaving. What still earns a click: things the answer can't do for them. Try the tool, see the pricing calculator, copy the full code, read the benchmark method, sign up. A page whose value is a fact that fits in one sentence will be summarized. A page whose value is doing something gets visited.

How to show up in AI Mode: steps for small sites

1. Get the page indexed first

Nothing else matters for a page Google hasn't indexed. Run URL Inspection on the pages you care about. If you see "Discovered" or "Crawled, currently not indexed", fix that before writing anything new (see Discovered and crawled, currently not indexed). On a new site, internal links from your homepage and a correct sitemap do more than any AI-specific tweak.

2. Put the content in the server HTML

Google says it can process JavaScript as long as it isn't blocked, and points to its JavaScript SEO practices for script-heavy sites. The safe choice is still to server-render the main content: the answer, the pricing, the comparison table. If curl on your URL returns an empty shell, fix that first. The free SEO checker shows what the server sends for one URL, plus title, canonical, robots and structured data.

3. Answer first, under every heading

Fan-out pulls in sections, not just pages. Open each H2 with one or two sentences that answer it on their own, then explain. Google says you don't need to write in a special way for AI, and you don't: this is what skimming readers want anyway.

4. Cover the follow-up questions

Think in conversations, not keywords. For each important page, write down the question someone asks first, then the three follow-ups they ask next. Good sources: your support inbox, sales calls, "People also ask", and long queries in your Search Console Performance report. Then make sure the page (or a clearly linked page) answers each follow-up with a heading and a direct answer.

One page, planned around a conversation (fictional product: an invoicing API)
TurnWhat they askWhere you answer it
FirstCheapest invoicing API for a small SaaS?H1 and first paragraph: who it's for, the price
Follow-upDoes it handle EU VAT?H2 “EU VAT”, answer in sentence one
Follow-upHow long does setup take?H2 with real steps and a code sample
Follow-upWhat happens if I outgrow the free plan?Pricing page, linked from the guide

Don't turn this into dozens of thin pages, one per guessed sub-query. Google's guide warns against creating pages for fan-out queries mainly to manipulate rankings or AI answers, and says its systems understand pages that cover several topics. One strong page with clear sections beats ten near-duplicates.

5. Publish what only you know

Google's May 2026 guide is blunt: optimizing for generative AI search "is optimizing for the search experience, and thus still SEO", and unique, non-commodity content is likely to matter more in the long run than any of its other suggestions. Indie builders have more of this than they think: your real usage numbers, the setup bug you hit and fixed, a benchmark on your own infrastructure, the price you actually paid. Date it and say how you measured it. A summary needs a source, and the source is whoever has the fact first.

6. Be a name the answer can mention

AI Mode answers name more products and companies than AI Overviews do (Ahrefs counted 3.3 entities on average versus 1.3). Make sure Google can connect your pages to one clear entity: the same product name everywhere, an About page, author names, and Organization markup with sameAs links to your real profiles. Real mentions help (a docs page that links your library, a newsletter that reviews your tool, a comparison that lists you). Bought or faked ones don't: Google's guide says inauthentic mentions "aren't as helpful as it might seem".

7. Hand the checks to your coding agent

Most of the above is code work. A prompt that works in Claude Code, Codex or Cursor:

Prompt: AI Mode eligibility and follow-ups
For each URL in app/sitemap.ts, curl the production page and report:
X-Robots-Tag, robots meta, any nosnippet / max-snippet / data-nosnippet,
and whether the H1 and the first paragraph under each H2 are in the server HTML.
Then, for /pricing and /docs/getting-started, list the H2s and say which of these
follow-up questions has no direct answer on the page: <paste 5 questions>.
Don't change anything and don't invent facts or numbers.

How to measure AI Mode visibility

Search Console now has first-party data, with limits that matter more on a low-traffic site. Google announced the Search generative AI performance report on June 3, 2026, and it reached all sites worldwide by August 31, 2026.

  1. 1Generative AI reportImpressions in AI Overviews and AI Mode, by page, country, device and date.
  2. 2Performance (Web)Clicks from AI features, mixed in with classic results.
  3. 3Follow-upsEach AI Mode follow-up is logged as a new query.
  4. 4Your own checksWhich questions cite you, and who else is cited.
Where AI Mode data shows up in Search Console. Impressions are split out; clicks are not.
  • The Generative AI report shows impressions only: how often links to your site appeared in AI Overviews or AI Mode, by page, country, device and date. There is no query dimension and no clicks, and AI Overviews and AI Mode are combined. Search Labs experiments aren't included.
  • Small sites may not see it at all. Google says the report may not appear if your site hasn't had enough impressions in generative AI features. That is normal early on, not a penalty.
  • Clicks stay in the main Performance report under the Web search type, together with classic results. A click on an external link in AI Mode counts as a click, and position follows the same method as a normal results page.
  • Follow-ups are new queries. Google says a follow-up in AI Mode is essentially a new query, with its impressions, position and clicks credited to it. You can't filter these from classic queries, but long, conversational queries in your report are worth reading as a list of follow-ups people ask.

Watch trends, not days. Compare month over month by page, note when you shipped a change, and remember that confirmed Google updates move results for everyone. For setup and permissions, see connecting Search Console.

  • Key pages are indexed (URL Inspection) and have no nosnippet, max-snippet:0 or noindex.
  • Settings > Search generative AI isn't set to exclude the property.
  • The main answer, tables and prices are in the server HTML.
  • Each H2 opens with a direct answer, and the common follow-ups have their own sections.
  • Each important page has something first-hand: data, a worked example, real numbers.
  • You review the Generative AI report and long queries in Performance once a month.

Where Serpkin fits

Serpkin doesn't track AI Mode answers. What it does check, monthly for your important queries, is whether Google shows an AI Overview, whether your site is cited and how many other domains are; when you aren't cited, it opens a "Get cited" mission your coding agent can pick up. Given the low overlap between the two features, treat that as a separate signal, not a proxy for AI Mode.

The part that helps most with the steps above is timing. Each mission shows a "what to expect" window, and with Search Console connected, Serpkin tracks when Google re-crawled the changed page, indexed it and first showed it in search. Its Search Console chart also shades confirmed Google update windows, so a drop during a core update doesn't get blamed on your last change. You can start a 7-day free trial. No citation or traffic change is guaranteed.

Check a page nowFree: see what a crawler finds on any URL, with a prompt for your agent.

Questions

How do I get my website to show up in Google AI Mode?

Make sure the page is indexed (check URL Inspection), eligible for a snippet (no nosnippet, max-snippet:0 or noindex) and not excluded in Search Console's Search generative AI setting. Then make it worth citing: direct answers under each heading, sections for common follow-up questions, and first-hand data others don't have.

Is Google AI Mode the same as AI Overviews?

No. AI Overviews are short summaries on some normal results pages. AI Mode is a separate conversational view with longer answers and follow-up questions. Google says they may use different models and techniques, and an Ahrefs study found their cited URLs overlapped only 13.7% of the time.

Can I see AI Mode traffic in Search Console?

Partly. The Search generative AI performance report shows impressions in AI Overviews and AI Mode combined, by page, country, device and date. It has no clicks or queries. Clicks from AI Mode are counted in the main Performance report under Web, mixed with classic results.

Do I need special schema or llms.txt for AI Mode?

No. Google says you don't need new machine-readable files, AI text files or special structured data to appear in its AI features. Structured data is still useful for rich results and describing your organization, but it isn't an AI Mode requirement.

How do I opt out of Google AI Mode?

In Search Console, go to Settings > Search generative AI and exclude the property. That removes its content from AI Overviews, AI Mode and Discover's generative AI features, usually within one to two days. Google says it isn't a ranking signal for the rest of Search and doesn't control AI training, which is Google-Extended.

Sources

Let your agent do the SEO work.

Serpkin finds the next fix from real search data, hands it to Claude Code, Codex or Cursor, and checks it after you ship.

npx serpkin
Start free