AI search news ·
AI can describe your brand accurately and still never bring it up: 96% recognised, 89% never mentioned
Victorious published its Q2 2026 Quarterly Search Report on July 29, 2026. Across 175 brands in legal, healthcare, SaaS, financial services and ecommerce, tested on eight AI platforms, the engines described 96% of brands accurately when asked about them by name — and 89% of the brands measured never appeared at all in the answers to category research questions. Recognition and recommendation turn out to be two different things, and only one of them is worth money. We have been measuring the same gap in a completely different population since June, and got the same shape.
Primary source: Victorious — Q2 2026 Quarterly Search Report
What the report actually measured
The primary source is Victorious' own Quarterly Search Report, Q2 2026 edition, covered by Search Engine Journal on July 29, 2026. The study ran two separate tests against the same brands, which is the design decision that makes it useful:
- A recognition test. Ask each platform to describe the brand by name, then grade the answer against the brand's own website on a five-point scale — correct, vague, outdated, wrong, or not recognised. 140 of the 175 brands produced evaluable responses. Result: 96% described accurately.
- A mention test. Ask the category research question a buyer would actually ask, and count whether the brand appears anywhere in the answer. 150 brands in this cohort. Result: 89% never appeared.
- Eight platforms, five verticals. ChatGPT, Claude, Gemini, Copilot, Perplexity, Google AI Overviews, Google AI Mode and Meta AI, across legal, healthcare, SaaS, financial services and ecommerce/retail.
Recognition accuracy is not uniform across engines. Victorious reports that Google AI Mode, Gemini, ChatGPT, Google AI Overviews and Copilot each cleared 83% accuracy in every vertical tested, while Perplexity recognised fewer than 55% of SaaS and ecommerce brands and Meta AI recognised 46% of SaaS brands. If your only evidence that "AI knows us" is a screenshot from one chatbot, that screenshot is a per-engine result, not a fact about the web.
We measured the same gap in a different population
Victorious tested mid-market brands with agency budgets. Every audit we run tests small local businesses, and it asks the engines the same two ways on purpose — once by name, four times the way a customer asks. Across 168 local businesses in 41 US metros (full dataset), the engines recognised the business in 98.8% of by-name answers and named it in 19.8% of buyer-question answers. 58% of those businesses appeared in none of their four buyer answers.
| Victorious Q2 2026 | AskedAbout (2026-07-21) | |
|---|---|---|
| Population | 175 brands, 5 verticals | 168 local businesses, 41 US metros |
| Engines | 8 platforms | ChatGPT + Perplexity |
| Described correctly when named | 96% | 98.8% |
| Appears in the buyer's question | 11% (89% never) | 19.8% (58% never in any) |
Two independent studies, different populations, different engine mixes, different question sets, one result: the models are not ignorant of these businesses — they simply do not volunteer them. That matters because the most common self-diagnosis we hear is the wrong one. "The AI doesn't know us yet, we need more content about us" is a plausible story that this data does not support. The model already knows you. It picked someone else.
Three things to do differently because of this
- Stop testing your brand by name. A by-name prompt is the one question where nearly everybody passes — 96% in Victorious' cohort, 98.8% in ours. It is a test that cannot fail informatively, and it is the test almost every brand runs on itself. Measure the category question instead: "who do you recommend for X in Y?" That is the number with revenue attached.
- Treat off-site mentions as the lever, but not as a guarantee. Victorious found referring domains correlate with AI mentions at r = 0.49 and third-party web mentions at r = 0.45 — real relationships, but nowhere near deterministic. The starker figure is the floor: brands with fewer than 2,000 indexed web pages mentioning them appeared in AI answers just 3% of the time. Below a certain volume of third-party presence, the engines have nothing to pick you from.
- Measure per engine, not "in AI". A 46% recognition rate on Meta AI and a 100% rate on ChatGPT for the same brand is not noise; it is two different indexes with two different opinions of you. Averaging them into one "AI visibility score" hides the engine where you are actually invisible.
The takeaway
The gap between known and recommended is now the most-replicated finding in this field, and it is the only one that changes what you should work on. If you want the number for a specific business rather than a cohort average, the free 60-second check asks ChatGPT and Perplexity the buyer's question and reports how often the business is actually named (the $79 audit adds Gemini and Claude, 25 questions sampled three times each). If you run this for clients, the agency 5-pack does it across five businesses at once. And if the answer comes back low, why AI recommends your competitors instead is the mechanism, not the mystery.
See your number
A free 60-second check shows what AI says about you.
Running this for clients? The $249 agency 5-pack audits five businesses, white-labeled.
Who runs this
- Built and operated by Sensara LLC, Atlanta, Georgia — about us and how the audit works.
- See what the report looks like before you run anything — score per engine, the competitors AI names instead of you, and a fix plan.
- We run the same audit on ourselves every week and publish the result: in the latest run AI named AskedAbout in 1 of 144 answers. We report our own numbers the way we report yours.