What is Google AI Mode and how does it choose sources?

Q: How is AI Mode different from AI Overviews?

AI Overviews are short AI-generated summaries that appear at the top of an otherwise normal results page. AI Mode is a separate chat-style destination where the whole page is the AI answer and you can ask follow-up questions that keep context. Both draw from the same Google index.

Q: Does traditional SEO still matter for Google AI Mode?

Yes. AI Mode reads the same index as classic Google search. If a page is not crawled, indexed, and ranking for a query, it has almost no path into the AI Mode answer. Traditional SEO earns the entry ticket, and GEO tactics like clear structure and schema decide whether the page gets quoted.

Q: What is query fan-out in AI Mode?

Query fan-out is the technique where AI Mode breaks your single question into several related sub-queries, runs each against the index, and synthesizes the results into one answer. That is why a single AI Mode answer can cite sources that each satisfy a different slice of your question.

Q: Does ranking first in Google guarantee a citation in AI Mode?

No. Ranking is the entry ticket, but the generative layer rewards the passage that answers a sub-query cleanly. Pages on page one of blue links can be skipped while a mid-ranked page with a crisp, directly-quotable answer gets cited instead.

Question

What is Google AI Mode and how does it choose sources?

Maciej Grabek · Accepted Answer

Google AI Mode is a conversational search experience that generates a written answer instead of a list of blue links. It runs on Google's existing search index and ranking signals, so it selects sources the way ranked search does: relevance to the query, freshness, site authority, and clear structure. It sits alongside AI Overviews and classic organic results, not in place of them. What Google AI Mode actually is AI Mode is Google's full conversational search surface. You ask a question in plain language, it returns a synthesized written answer with a handful of cited links, and you can ask follow-up questions that keep the context. It is a step past AI Overviews, which are the shorter AI-generated summaries that appear at the top of an otherwise normal results page. Think of it as three layers of the same product. Classic organic results are the ranked ten blue links. AI Overviews are a summary bolted on top of those results for some queries. AI Mode is a separate, chat-style destination where the whole page is the AI answer. All three draw from the same underlying Google index. How AI Mode picks and cites sources The mechanism Google has described is query fan-out. AI Mode takes your one question, breaks it into several related sub-queries, runs each against the index, and then synthesizes the results into one answer. That means a single page can end up citing sources that satisfy different slices of your question. The signals that decide which pages get pulled in are the familiar ranking ones, applied to a generative context: Answer relevance: does the page directly address the sub-query, not just contain the keywords. Freshness: for anything time-sensitive, recently updated pages get preference. Authority and trust: established sites and pages with real expertise are pulled more often, the same E-E-A-T logic Google already uses. Structured data: clear headings, direct question-and-answer formatting, and schema markup make a page easier to lift a quotable passage from. Citations show up as inline links or a source panel next to the answer. Being cited is not the same as ranking first in classic results. We see pages that sit on page one of blue links get skipped in AI Mode, and mid-ranked pages with a crisp, directly-quotable answer get cited instead. The generative layer rewards the passage that answers the sub-query cleanly, not only the domain with the most backlinks. Why it matters that Google's index still underpins it This is the practical takeaway that gets missed. AI Mode is not a separate web that you optimize for with special tricks. It reads the same index that classic Google search reads. If your page is not crawled, indexed, and ranking for a query, it has almost no path into the AI Mode answer for that query. So traditional SEO and GEO are not competing strategies here, they compound. Getting indexed, earning authority, and ranking is the entry ticket. Then generative-engine optimization decides whether you get lifted into the answer: a clean TL;DR near the top, question-shaped headings, direct declarative sentences, and structured data all raise the odds that AI Mode quotes you rather than a competitor covering the same ground. A concrete example Suppose you ask AI Mode: best project management tool for a small remote team AI Mode does not just find one page titled that. It fans the question out into sub-queries like "project management tools for small teams," "best tools for remote collaboration," and "project management pricing for startups." It then writes a paragraph naming a few options, usually surfacing familiar names like Asana, Trello, and ClickUp, and cites four to six links beside the answer: a comparison roundup, a couple of the vendors' own feature pages, and often a review site like G2. The vendor that wrote a clear, structured comparison page is the one most likely to be cited, even if a thinner competitor outranks it in classic blue links. An honest caveat This space is moving fast. Google has changed how and where AI Mode and AI Overviews appear, which queries trigger them, and how prominently sources are shown, more than once. Treat the specific mechanics as a snapshot, not a permanent contract. The durable part is the foundation: Google's index and ranking signals still decide what AI Mode can even see, and clear, well-structured, genuinely useful pages are the ones that get cited. Optimize for that and you stay resilient as the surface shifts. To know whether AI Mode is actually citing you, you have to check. Run your real buyer queries in AI Mode and note which pages get cited and which competitors show up. Doing this by hand across a set of prompts and engines gets tedious, which is where a monitoring tool earns its place: avisibli tracks which sources ChatGPT, Perplexity, Gemini, Claude, and DeepSeek cite for your prompts over time, so you can see movement instead of spot-checking once and guessing. avisibli is the GEO platform that publishes this answer library. Sel...