A buyer asks an AI assistant how to fix a problem, and the answer cites a handful of sources. Weeks later the same buyer asks which company to hire, and the answer names a handful of vendors. A new report says those two lists overlap far more than you would expect if they were picked separately.
Victorious, a San Francisco search and AEO agency, published its Q3 2026 Search Report on October 25. The headline number is stark. Of the brands cited on research questions, 94% were also named on buying questions; of the brands with no research citation, 50% were.
What the report measured
The base is 161 brands in five industries and about 16,500 AI answers from nine platforms. The answers come from Ahrefs Brand Radar, which tracks the questions US searchers put to AI tools, and Victorious sorted them into research questions (learning about a problem) and buying questions (choosing a company).
Some things are missing. The release does not name the nine platforms, the five industries or any of the brands, and points to Victorious's site for the full method and sample sizes. Keep that in mind below.
The research-to-buying link
Here is the core claim. Victorious says the 94% versus 50% gap held even after it controlled for how often AI mentions each brand overall, which means this is not simply a case of famous brands turning up in every answer.
The sharper number tracks movement. On July 1, 40 brands in the sample had not been named on a single buying question. By October, 81% of those with a research-stage citation had been named on at least one. Without one? 21%.
CEO Michael Transon put it this way: research content "has been treated as a traffic play," but in AI search "it looks more like a qualifying round."
The lists are moving, and getting harder to enter
Churn first. Of the 42 brands AI named in Q2, 15 were gone by Q3, and only two new ones came in. A spot in the answer is rented, not owned.
The bar is also rising. Victorious estimates a brand now needs roughly 24,000 to 28,000 third-party web mentions for an even chance of being named, up from about 20,000, and in every industry it studied the ten most-cited websites took a larger share of AI citations, reaching about half in legal services.
Then there is reach. Google's AI Overviews now appear on 97.5% of the category searches Victorious tracked, up from 72.8%. For these categories the AI answer is simply the first thing a buyer reads.
Where we would push back
It is a correlation, and the release says so. A company that writes strong explainer content often also has a big sales team, a long page of reviews on G2 and steady press coverage, and any of those could be what gets it named when buyers ask who to choose. Controlling for overall mention volume helps. It does not settle the question.
Other limits. The 40-brand subgroup behind the 81% versus 21% figure is small, and the data is US-only. Victorious sells AI search visibility work, so it has a stake in research-stage content paying off (so do we, which is why we read the method before the headline). EIN Presswire, which carried the release, does not check the claims it distributes.
Still, a gap of 94% against 50% is wide, and it survived the one control that matters most. Read it as a strong hint about where to look first, not as proof of cause.
What to do with it
For a B2B company the practical reading is short. Stop judging explainer and guide pages only by the clicks they bring in, and check instead whether ChatGPT, Claude and Google AI Overviews cite them when a buyer asks how the problem works. If they do, you have a better shot at being named when that buyer asks who to hire. If they don't, that is probably the gap to close before you touch any page that pitches your product.
Then remember the churn. Fifteen of 42 brands dropped out in one quarter. Being named once is a reading, not a result, so measure every week and treat a drop as a signal.
The bigger picture
AI search has two stages, and this report suggests the first one feeds the second. The engines settle early on which sources they trust for a topic, and that trust carries into the moment a buyer is ready to pick a vendor. Invisible during research means starting the buying question from behind.
ClawWorld runs AI visibility work for B2B startups. Start with a free baseline report: 10 buyer questions, who ChatGPT and Claude name instead of you, and the pages they cite.