You rank in AI Overviews. So you assume you are covered in AI Mode too.
You are not.
Ahrefs looked at 730,000 AI responses and found that AI Mode and AI Overviews cite the same URL only 13.7% of the time. Same query, same Google, two answers pulling from almost entirely different pages. Winning one tells you next to nothing about the other.
This trips up a lot of good sites. They see their name in the AI Overview, assume the job is done, and never notice they are missing from the surface where buyers ask longer, higher-intent questions. Different game, different pages, different work.
AI Mode and AI Overviews are not the same product
Separate the two first, because people say the names like they mean the same thing.
AI Overviews is the summary box that sits on top of your normal search results. You did not ask for it. Google built it from the pages already ranking for your query, and it sits above the ten blue links.
AI Mode is a place you go on purpose. It is a chat-style search, run by Gemini, where you type a real question and get a composed answer with sources. It is closer to Perplexity than to a classic results page, and the same signals that get you cited by Gemini are in play here.
Two products. Two retrieval systems. That is the root of everything below.
How little the two share
The overlap is smaller than most people guess.
Ahrefs put the shared-URL rate at 13.7% across those 730,000 responses. Victorious ran its own study and found only 30 to 35% of the URLs in AI Overviews also showed up in AI Mode, and that 77% of domains appear on just one of the two surfaces, never both.
Read that last number again. Most sites that earn a spot earn it on one surface only. The idea that ranking flows straight from AI Overviews into AI Mode is wrong, and the data is not close.
The strange part: the two usually agree on the answer. Victorious measured about 86% semantic similarity between the responses. Same conclusion, different receipts. Google reaches the same take from two different piles of sources.
Why they pull different URLs: query fan-out
The mechanism behind the split has a name: query fan-out.
When you ask AI Mode something, it does not run one search. It breaks your question into a set of sub-queries, runs them in parallel, and pulls sources for each. Aleyda Solis and iPullRank have both documented how this works: one query becomes many, each chasing a different angle, freshness signal, or entity.
Then it ranks. AI Mode scores pages by how well they cover the whole expanded set of sub-queries, not just your original keyword. A page that answers one narrow question well can lose to a page that touches six of the sub-questions.
A quick example. Search “best CRM for a small sales team” in AI Mode and it does not just hunt for that phrase. It fans out: pricing for small teams, ease of setup, integrations, the specific tools people compare it against, reviews from teams that size. Each branch pulls its own sources. The page cited is rarely the one ranking first for the exact phrase. It is the one that answers four or five of those branches in a single place.
That is why the URLs differ. AI Overviews leans on the pages already ranking for the head query. AI Mode’s fan-out reaches deeper and wider, into pages that never sat in the top ranking positions. It draws from the live index, the Knowledge Graph, structured data, even the Shopping Graph. Different retrieval, different sources, small overlap. Getting named across both is what generative engine optimization is built for.
Same answer, different sources
One thing here changes the strategy.
Google is not disagreeing with itself. The 86% semantic similarity Victorious found means the two surfaces mostly say the same thing. They differ on which pages they credit for it.
So the goal is not to change the answer. The answer is settled. The goal is to be one of the pages Google reaches for when it builds that answer in AI Mode, the same way you work to be one it reaches for when optimizing for AI Overviews. Two lists to get on, one message to support.
How to optimize for AI Mode
AI Overviews rewards the clean, concise answer. AI Mode rewards the page that covers the whole question. In practice, that means a few things.
Answer the sub-questions, not just the keyword. Query fan-out means AI Mode is really asking six or eight things at once. A page that only nails the head term shows up for one of them. Map the questions a searcher asks before and after the main one, and answer each on the page, in its own clear section.
Go deep and multi-entity. Victorious found AI Mode favors comparison-rich, multi-entity content over the short snippet. Name the alternatives, the trade-offs, the specifics. Thin pages get filtered out of the deeper fan-out.
Make your entity legible. AI Mode pulls from the Knowledge Graph and structured data, so a clear brand entity gets you into more of the sub-queries. Clean structured data does the same, feeding the model facts it can trust. This is the same foundation the rest of your answer engine optimization runs on.
A worked contrast makes it concrete. A pricing page with one boxed answer, “plans start at $49 a month,” is built for AI Overviews. A page that lays out the tiers, compares them to two named competitors, explains who each tier fits, and answers the questions a buyer asks next is built for AI Mode. Same topic, two shapes. The strongest pages carry both: the clean answer near the top, the depth underneath it.
None of this is a trick. Cover the question, prove the entity, and let the fan-out find you in more than one place.
How to check whether AI Mode cites you
You do not need a tool to start. You need to look.
Open AI Mode and ask the questions your buyers ask. Read the sources it credits. Run the same questions through AI Overviews and compare the two lists. Where you show up in one and not the other, you have found your next piece of work.
Do this for your top ten buyer questions and you have a map: where you are cited, where a competitor is cited in your place, and which surface is leaking. That map is worth more than any visibility score, because it is your real answers, on your real queries, today.
Where to start
Two surfaces, one message, different pages behind each. That is the whole thing.
Pick your ten highest-intent questions. Check both surfaces. Then build pages that answer the full question, not just the keyword, with the entity and schema that let the fan-out reach you. AI Overviews will reward the concise answer inside those pages. AI Mode will reward the depth around it.
If you would rather not run two optimization tracks yourself, that is the work I do. Bring it to an AI SEO specialist who tracks both surfaces every week, and you skip the guesswork about which one is dropping you.
Sources: AI Overviews vs. AI Mode, Ahrefs. New Study: Google AI Overviews vs. AI Mode, Victorious via DesignRush. Google AI Mode’s Query Fan-Out Technique, Aleyda Solis. Expanding Queries With Fan-Out, iPullRank.
Common questions,
direct answers.
01 Is Google AI Mode the same as AI Overviews?
No. AI Overviews is the summary that sits above your normal search results, generated from pages already ranking for your query. AI Mode is a separate, chat-style search you open on purpose, run by Gemini. They behave differently and cite different pages. Ahrefs found the two share the same URL only 13.7% of the time.
02 Why does AI Mode cite different URLs than AI Overviews?
Query fan-out. AI Mode breaks your search into many sub-queries, runs them at once, and pulls sources from a wider pool than the top ranking results. It then ranks pages by how well they cover the whole expanded set. That reaches URLs the ranking results never surface, which is why the overlap with AI Overviews is small even though the two usually reach the same answer.
03 If I rank in AI Overviews, will I show up in AI Mode?
Not automatically. Victorious found that 77% of domains appear on only one surface, not both. A page built to win the concise AI Overviews snippet can be invisible in AI Mode, which favors deeper, multi-entity content. You have to earn each surface on its own terms.
04 How do I optimize for Google AI Mode?
Cover the whole question, not just the head keyword. Answer the sub-questions a searcher asks next, in depth, with clear entities and structured data the model can read. AI Mode rewards comparison-rich, multi-entity pages over a single short answer. Clean schema and a strong brand entity help you get pulled into more of the fan-out queries.
05 How do I check whether AI Mode cites my site?
Open AI Mode, ask the questions your buyers ask, and read the sources it lists. Run the same questions through AI Overviews and compare. If you show up in one and not the other, that gap tells you which surface to work on next. Track it over time, because the sources shift as content updates.