How to Rank in Google AI Mode: 8-Step Plan

Short answerRanking in Google AI Mode means being cited as a source or named as a brand in its answers. AI Mode splits each question into subtopics, searches each one, and writes from the pages it finds. To rank, your pages must be indexed and snippet-eligible, answer the sub-questions and follow-ups buyers ask, and be named on the pages Google retrieves.

Key takeaways

Google says there are "no additional requirements" for AI Mode: index the page, allow a snippet, and skip the special schema and AI text files.
AI Mode and AI Overviews cite different pages: Ahrefs measured only 13.7% citation overlap across 540,000 query pairs, so ranking in one doesn't mean ranking in the other.
Only 14% of AI Mode URLs matched Google's organic top 10 for the same keyword in SE Ranking's 10,000-keyword study, and just 9.2% of URLs repeated across three same-day runs.
AI Mode names more brands than AI Overviews: 3.3 entities per answer against 1.3, and Wikipedia made up 28.9% of its citations in Ahrefs' data.
Search Console counts AI Mode clicks, impressions and positions inside the Performance report's Web search type, and its Generative AI report shows impressions only.

What does it mean to rank in Google AI Mode?

AI Mode is Google's conversational search tab, where you ask a full question, get a long written answer with links, and keep asking follow-ups. To rank there is to be one of the links in that answer, or one of the brands the answer names, for the questions your buyers ask. There's no numbered list of ten results to climb.

It's now a mass channel. Google said at I/O 2026 that "AI Mode has surpassed one billion monthly users," with queries "more than doubling every quarter since launch"1 (Google). It's available in over 200 countries and territories2, and since I/O 2026 people can ask a follow-up straight from an AI Overview and "flow into a conversational back and forth with AI Mode"1. That last change matters for planning: a buyer who starts in an AI Overview can end up in AI Mode without ever choosing it.

This guide is about AI Mode specifically. If you want the shorter summaries at the top of normal results, those have their own AI Overviews guide. The two share an index but behave differently enough to plan for separately:

	AI Overviews	AI Mode
Where it shows	Above regular results, when Google decides it helps	Its own tab and conversation; also reached by follow-ups from an AI Overview
Answer length	Short summary	About 4x longer (Ahrefs)
Brands and entities named	1.3 per answer on average	3.3 per answer on average
Answers with no citation	11%	3%
Wikipedia share of citations	18.1%	28.9%
Follow-up questions	Hand off into AI Mode	Each one counts as a new query in Search Console

Comparison figures are from Ahrefs' study of US data from September 20258 (Ahrefs). The same study found the two give semantically similar answers (86% average similarity) while citing different pages, overlapping on just 13.7% of citations.

How does Google AI Mode choose its sources?

Query fan-out: many searches behind one answer

Google describes AI Mode as "breaking down your question into subtopics and issuing a multitude of queries simultaneously on your behalf"3 (Google). Search Console's help page puts it the same way: AI Mode "groups the user's question into subtopics and searches for each one simultaneously"5. Deep Search, the research setting, "can issue hundreds of searches"3. Google hasn't published how many sub-searches a normal answer runs, so treat any specific number you read as a vendor estimate.

This is query fan-out, and it's why a page can be cited for a question it doesn't rank for. AI Mode isn't ranking your page against the buyer's sentence. It's ranking pages for each sub-search, then writing from the winners. A page that ranks well for one narrow sub-question ("does X integrate with NetSuite") can be cited in an answer to a broad one ("best expense software for a 300-person company").

Eligibility: the same rules as Search

Google's documentation for site owners is blunt. 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," and "there are no additional technical requirements." You don't need "new machine readable files, AI text files, or markup," and there's "no special schema.org structured data that you need to add"4 (Google Search Central). The controls are the normal ones: robots.txt for Googlebot, and nosnippet, data-nosnippet, max-snippet or noindex to limit what's shown.

Different models, different pages

Google also says AI Mode and AI Overviews "may use different models and techniques, so the set of responses and links they show will vary"4. AI Mode now runs on Gemini 3.5 Flash by default1. In practice, three numbers tell you how loosely AI Mode follows the regular rankings:

SE Ranking found 14% URL overlap between AI Mode citations and the organic top 10 across 9,734 AI Mode answers, with 12.6 URLs per answer9.
seoClarity found that 81% of 1,000 transactional queries had at least one AI Mode citation from the organic top 20, but only 28% had one from the top 10. When a page did rank first, it was cited 25% of the time10.
Answers are unstable. Only 9.2% of URLs appeared in all three of SE Ranking's same-day runs for the same keyword9.

A top-three ranking helps. So does ranking 11th to 20th for the right sub-question.

Our takeAI Mode rewards being the best page for a narrow question more than being the strongest domain for a broad one. The pages it cites are usually answers to the sub-questions it invented, and the brands it names are the ones those pages agree on. That makes AI Mode closer to a long-tail SEO problem than to a branding one: map the sub-questions, win them one page at a time, and get named on the pages that win the rest.

The 8-step plan to rank in Google AI Mode

Steps 1 to 3 take a few days and steps 4 to 8 are the ongoing work. The table is the checklist: don't move on until the "done when" test passes. The detail for each step follows it.

Step	Done when	Time
1. Write prompts the way buyers talk to AI Mode	You have at least 30 first questions, each with one written follow-up, and no more than 20% of them mention your brand.	Half a day
2. Run a baseline three times	You have a sheet with 90+ runs, your mention rate and citation rate per question, and the list of every cited URL with how many questions it appeared on.	Half a day to a day
3. Check that your key pages are eligible	Every money page shows "URL is on Google," returns 200 to Googlebot, and has no snippet restriction you didn't choose on purpose.	An hour
4. Map the sub-questions behind each prompt	Each of your top 15 prompts has a list of sub-questions, and each sub-question has either "we own the cited page," "a third party owns it" or "nobody clearly owns it."	A day for 15 prompts
5. Give every sub-question you can own its own answer	Every sub-question you can credibly answer has a section that answers it in two sentences, with at least one fact (number, price, named integration) a competitor page doesn't have.	Ongoing, about a page a week
6. Make your brand an entity Google can name	The same one-line category description and core facts appear word for word on your site and on every profile you control.	A day, then upkeep
7. Get named on the pages AI Mode cites	You're named on at least half of the 20 most-cited third-party URLs. Budget about a quarter for this, since it runs on other people's schedules.	About a quarter
8. Feed Google's other sources: Maps, Shopping and preferred sources	Your Business Profile and Merchant Center data (where relevant) match your site exactly, and publishers have the preferred sources button live.	A day

1. Write prompts the way buyers talk to AI Mode

AI Mode queries are longer than search queries, and Google says people ask "longer and more specific questions" in its AI experiences6. Write 30 to 50 buyer questions with a constraint in each ("for a restaurant group with 12 locations," "under $10 per employee"), and give each one the follow-up a buyer would ask next ("which of those integrates with Toast?"). Pull the wording from sales calls and from long queries in Search Console.

2. Run a baseline three times

Run every question and its follow-up in AI Mode three times, signed out or in a clean browser profile, from the country you sell in. Log the brands named, their order, and every cited URL. With only 9.2% of URLs repeating across three runs9, one run tells you almost nothing. Count a brand or URL as present for a question only if it appears in at least two of three runs.

3. Check that your key pages are eligible

Use URL Inspection in Search Console on your pricing, product, comparison and top guide pages. Every one of them should show as indexed, with no nosnippet tag and no restrictive max-snippet value, because Google says "more restrictive permissions will limit how your content is featured in our AI experiences"6. Check your CDN or firewall isn't challenging Googlebot. Ignore Google-Extended for this job: Google says it "does not impact a site's inclusion in Google Search"11, which covers AI Mode. Our AI crawler guide explains the difference.

4. Map the sub-questions behind each prompt

For each prompt, write the 5 to 10 searches you think AI Mode runs: the category, the constraint, pricing, integrations, alternatives, reviews, and the follow-up. Then check the cited URLs from step 2 against that list; each cited page usually answers one sub-question. That gives you a grid of sub-questions with the page currently winning each.

5. Give every sub-question you can own its own answer

For the "nobody owns it" rows, build or fix the page. Put the direct answer in the first two sentences under a heading that matches the sub-question, then the detail: a table, prices, limits, an example. AI Mode answers are long and name several brands, so pages that compare (with a clear "choose X if" line) and pages that state hard facts (price, plan limits, supported integrations) get used. Google's own advice is to focus on "unique, non-commodity content"6; in practice that means the numbers and specifics only you can publish. Run drafts through the citability grader before publishing.

6. Make your brand an entity Google can name

AI Mode names 3.3 entities per answer on average, against 1.3 in AI Overviews, and leans on Wikipedia for 28.9% of citations8. It names brands it can identify with confidence. Make the basic facts about you (category, founding year, headquarters, pricing model, main integrations) match across your About page, LinkedIn, Crunchbase, review profiles and, if you qualify, Wikipedia and Wikidata. Add Organization schema with sameAs links; the schema generator builds it. Google says no special schema is needed for AI features, so treat this as consistency work.

7. Get named on the pages AI Mode cites

Work the "a third party owns it" rows. These are usually review sites, roundups, Reddit threads, YouTube videos and, for jobs and finance topics, sites like Indeed and NerdWallet, which were among the most-cited domains in SE Ranking's data9. Pitch roundup authors with something checkable, refresh your review profiles in the categories AI Mode pulled, and answer cited threads where you have something specific to add. Prioritize pages that appeared on the most prompts in step 2.

8. Feed Google's other sources: Maps, Shopping and preferred sources

AI Mode draws on more than web pages. In SE Ranking's study, google.com links made up 5.7% of citations, and 97.9% of those pointed to Google Maps business profiles9. For products, AI Mode uses the Shopping Graph, which Google says holds "more than 50 billion product listings," with over 2 billion refreshed every hour12. So: keep your Business Profile complete if you serve local customers, keep Merchant Center feeds accurate if you sell products, and if you publish content, add Google's preferred sources button, which can give your site a "preferred" badge in AI Mode for users who choose you13.

Worked example: a restaurant payroll tool

Here's the method applied to a hypothetical company, "Shiftpay," payroll software for restaurant groups. We haven't run these prompts; the sub-searches are our guesses at what fan-out would produce, which is exactly what step 4 asks you to write down.

One prompt and its follow-up.

Q1  what's the best payroll software for a restaurant group with 12 locations
    and tipped employees
Q1b which of those integrate with Toast POS

Likely sub-searches (our guess, not Google's):
  best restaurant payroll software
  payroll software multi-location restaurants
  tip pooling payroll software
  restaurant payroll software pricing per employee
  payroll software Toast integration
  [competitor A] vs [competitor B] restaurant payroll
  restaurant payroll software reviews

The grid after a baseline. Suppose three runs of Q1 and Q1b cite nine distinct URLs. Shiftpay's team sorts them by sub-question:

Sub-question	Cited page (hypothetical)	Owner	Shiftpay named?	Action
Best restaurant payroll	Review-site category page	Third party	No	Step 7: complete profile, ask customers for reviews
Tip pooling	Competitor's guide	Competitor	No	Step 5: publish a tip pooling page with worked numbers
Pricing per employee	Roundup with a price table	Third party	Yes, but old price	Step 6: send the author the current price page
Toast integration	Toast's integrations directory	Third party	No	Step 7: get listed in the directory
Reviews	Reddit thread	Third party	No	Step 7: answer only if there's something specific to add

Reading it. Rewriting Shiftpay's homepage would touch none of these rows. The fastest wins are the Toast directory listing (one form, and it answers the follow-up outright) and correcting the price in a roundup that already names Shiftpay. The tip pooling page is the one owned asset worth building, because nobody but a competitor answers that sub-question today.

What doesn't move AI Mode rankings

AI text files and special markup

Google says you don't need "AI text files" or special schema to appear in AI Mode4. An llms.txt file is harmless, but Google hasn't said it reads one for AI Mode, and its documentation says nothing extra is required.

Optimizing for AI Overviews and assuming it carries over

With 13.7% citation overlap between the two8, tracking only AI Overviews leaves most of AI Mode unmeasured. Run your prompts in both.

Blocking or allowing Google-Extended

It doesn't change your AI Mode eligibility either way11. If you want out of AI Mode, the only levers are the snippet controls and noindex, and they take you out of regular results too.

Expecting ads to buy the answer

Google is testing ad formats in AI Mode, and they're labeled "Sponsored"14. They sit beside the organic answer; they don't decide which brands it names.

Chasing the head term only

Only 28% of transactional queries in seoClarity's study had an AI Mode citation from the organic top 1010. The pages that get cited often answer a narrower sub-question. A plan that only targets "best [category] software" leaves most citations to whoever answers the specifics.

How do you measure AI Mode rankings?

What Search Console shows. Google says AI features "are included in the overall search traffic in Search Console," inside the Performance report's Web search type4. Its help page defines the AI Mode rules: "Clicking a link to an external page in AI Mode counts as a click," "standard impression rules apply," position "follows the same methodology as a Google Search results page," and a follow-up question is "essentially performing a new query"5 (Search Console Help).

The Generative AI report. Since June 2026, Search Console has a dedicated generative AI performance report, which Google finished rolling out to all websites on August 31, 2026. It shows impressions within features "such as AI Overviews and AI Mode," broken down by page, country, device and date7 (Google Search Central). It doesn't show clicks or queries, and we haven't found a filter that separates AI Mode from AI Overviews, so it tells you which pages surface in Google's AI features combined.

Why clicks undercount. Semrush's clickstream study of almost 69 million Google sessions found 92% to 94% of US desktop AI Mode searches ended without a visit to another site15. A brand named in an answer can influence a buyer who never clicks. That's why the prompt baseline from step 2 matters: re-run it monthly, and report mention rate, citation rate and share of voice alongside Search Console impressions.

By hand, 30 prompts with follow-ups x 3 runs is an afternoon. For weekly runs across countries, a tracker is cheaper than staff time; our AI Mode tracker comparison lists which tools cover AI Mode and on which plan. For a quick first read on whether AI engines name your brand at all, Semrush's free checker includes AI Mode; our checker roundup has the others. Our guide to measuring AI visibility has the sample-size math and a reporting template.

How quickly does AI Mode reflect changes?

Eligibility fixes land as soon as Google recrawls the page. New pages have to win their sub-question in regular ranking first, which takes weeks on a strong site and months on a new one. Third-party mentions move on other people's schedules. Give the plan a quarter, then judge it on mention rate across the whole prompt set.

How do you rank in the other AI engines?

Each engine retrieves differently. ChatGPT leans on its own crawler and third-party search (how to rank in ChatGPT). Perplexity runs its own index (how to rank in Perplexity). The Gemini app grounds in Google Search plus Google's other graphs (how to rank in Gemini). Claude searches the web through its own tooling (how to rank in Claude). For the shared mechanics, see how AI engines choose their sources, and for usage data across engines, our AI search statistics.

Frequently asked questions

Does AI Mode need different SEO work?

Eligibility works the same way for both: an indexed page that's allowed a snippet qualifies. What changes is the target. AI Mode runs many sub-searches per question and writes one answer, so you win by owning narrow sub-questions and being named on the third-party pages it retrieves, more than by ranking first for one head term.

Do I need schema or llms.txt to appear in AI Mode?

No. Google's documentation rules out any special schema.org type or AI text file as a requirement for either AI feature. Structured data should match visible content. Organization schema is still useful housekeeping because it keeps your brand facts consistent.

Can I see AI Mode traffic in Google Search Console?

Partly. AI Mode clicks, impressions and positions are included in the Performance report under the Web search type, mixed in with regular results. The Generative AI report adds impressions from AI features (AI Mode included) that you can split by date, device, country or page; it has no clicks or queries, and it doesn't separate AI Mode from AI Overviews.

Does ranking in AI Overviews mean I'll rank in AI Mode?

Not reliably. Ahrefs found the two cite the same pages only 13.7% of the time, even though their answers are about 86% similar in meaning. Google says they may use different models and techniques. Track your prompts in both, and plan for AI Mode's longer answers, which name more brands.

How do I opt out of Google AI Mode?

You can limit what AI Mode shows with nosnippet, data-nosnippet, max-snippet or noindex, the same controls that govern regular search snippets. Blocking Google-Extended doesn't work, because Google says it doesn't affect inclusion in Search. Any opt-out strong enough to remove you from AI Mode also reduces how you appear in normal results.

Why does AI Mode cite pages that don't rank on page one?

Because it runs many sub-searches behind each question. A page ranking 15th for your query might rank first for one of the sub-searches AI Mode runs behind the scenes. seoClarity found 81% of queries had a citation from the organic top 20 but only 28% from the top 10.

Sources

Cite this page

AI Ranked Editorial. "How to rank in Google AI Mode: an 8-step plan for getting cited and named." AI Ranked, October 11, 2026. https://airanked.ai/guides/how-to-rank-in-google-ai-mode