Google's query fan-out splits AI queries against classic search
Summary
Google's Liz Reid explained that AI Overviews break complex queries into smaller sub-queries and run each through classic search. Google calls this "query fan-out."
Your page doesn't need to match the full conversational question. It needs to win one of the sub-queries, which look a lot like traditional keyword phrases.
Don't build pages for one-off long-tail AI queries. Focus on being the best answer for specific, recurring sub-topics within your existing content.
What happened
Google’s Liz Reid explained on the Bloomberg Odd Lots podcast how AI Overviews and AI Mode are changing query behavior. Users are now expressing full, complex information needs instead of condensing them into short keyword phrases. Search Engine Journal’s coverage by Roger Montti unpacks the key implications for SEO.
Reid described how a user searching for “restaurants New York” always had a more complex need in mind. They wanted a restaurant in a specific location, for five people, not too pricey, with vegan options and kid-friendly seating. In the old model, users translated that need into “keyword-ese.” Now they type the full question and expect Google to do the translation.
The critical detail is what happens next. Google doesn’t match that long query against a single page. Instead, it uses query fan-out to decompose the complex question into smaller, specific sub-queries. Each sub-query runs against classic search. Google’s AI then selects from the results across those sub-queries and synthesizes an answer.
Why it matters
Query fan-out means the unit of ranking hasn’t changed as much as it might seem. Long, complex AI queries get broken into fragments that look a lot like traditional keyword phrases. Your page doesn’t need to answer the entire complex question. It needs to be the best result for one of the decomposed sub-queries.
The practical implication is that creating pages targeting hyper-specific long-tail AI queries may not be the right move. Those complex queries are often one-off and rarely repeated. The time spent crafting content for them could be better spent elsewhere.
Montti’s analysis points out that because AI Overviews share space among multiple sites, other factors gain importance. Relevant images and video content can help a site stand out within an AIO result.
A separate SEJ article on “browsy” queries adds context. Reid noted that user behavior across search surfaces is varied, not monolithic. Some queries still favor the full SERP experience, particularly what she calls “browsy” queries where users want to explore rather than get a single synthesized answer.
Meanwhile, Google’s Nikola Todorovic encouraged SEOs to use AI tools themselves to analyze data, research competition, and improve their ability to provide value. The message from Google’s side is consistent: focus on being genuinely useful for specific needs.
What to do
Audit your content against decomposed queries, not full complex questions. Think about what sub-queries your pages could satisfy when a complex AI question gets broken apart. A page about “kid-friendly vegan restaurants in Midtown Manhattan” answers a very specific fragment of a larger query.
Don’t chase one-off long-tail AI queries. If a complex query is unlikely to be repeated, building a page around it has poor ROI. Focus instead on the recurring, specific sub-queries that fan-out generates repeatedly across many different complex questions.
Strengthen your position in classic search for specific queries. Query fan-out runs against classic search results. The fundamentals of ranking for well-defined keyword phrases still apply. If you rank well for a sub-query, you’re a candidate for the synthesized AI answer.
Claim visual space in AI Overviews. Since AIO results pull from multiple sites, differentiation matters. Include relevant images and video on key pages so Google has rich media to surface alongside your content.
Track which sub-queries your content wins. Google Search Console won’t show you fan-out sub-queries directly, but monitoring which specific queries drive AI Overview impressions can reveal the fragments Google associates with your pages.
Watch out for
Misreading the signal as “long-tail is back.” Reid’s comments might tempt some practitioners to build content for verbose, conversational queries. The fan-out mechanism means Google decomposes those into shorter fragments. The winning pages answer specific sub-queries, not the full conversational prompt.
Assuming AI search replaces classic search behavior. Reid explicitly noted that search behavior is varied across surfaces. Some users still prefer browsing full SERPs. Treating AI Overviews as the only surface worth targeting risks neglecting traffic from traditional results.