87% of brands that appear in AI answers never get recommended

Across 270 live AI prompt panels, 247 brands were mentioned at least once. Only 31 of those 247 ever received a recommendation. The remaining 216 appeared in AI answers but never received the signal that drives purchase decisions: "you should use this one."

Mention and recommendation are not the same thing

A brand can appear in an AI answer in two very different ways. It can be mentioned as part of a category overview, a comparison, or a context note. Or it can be recommended as the answer to a buyer's question. The first says "this brand exists." The second says "use this one."

Most AI visibility measurement treats these as equivalent. A brand that shows up in 20% of queries looks like it has 20% visibility. The problem is that "showing up" and "getting the recommendation" are almost entirely different outcomes, driven by different factors, and producing very different downstream effects on buyer behavior.

We measured both for 270 B2B software brands across project management, customer support, and HR software categories. The gap between the two numbers is larger than most teams would expect.

The numbers

Outcome	Brands	Share of 270
Never mentioned	23	8.5%
Mentioned, never recommended	216	80.0%
Mentioned and recommended	31	11.5%

The 216 brands in the "mentioned, never recommended" group are not invisible. Their median AI mention rate is 21%, which is right at the field median for all 270 brands. These are brands that show up regularly in AI answers. They are just never the answer.

What the recommendation signal looks like in practice

When a buyer asks an AI assistant "what is the best customer support software for a 50-person team," the response does not list every brand it has heard of. It recommends two or three. The brands in the "mentioned" category typically appear in the supporting text: as alternatives, as comparisons, as footnotes. The brands in the "recommended" category are named first.

In the data, recommendations concentrate sharply. Five brands captured the majority of the 250+ recommendation signals across these panels:

Brand	Times recommended (across 270 panels)	Category
Asana	284	Project management
BambooHR	72	HR software
Jira	68	Project management
ClickUp	63	Project management
Zendesk	58	Customer support

Asana alone received 284 recommendation signals across 270 panels. A brand with a 21% mention rate received zero. Both brands "appeared in AI answers." Only one of them is being recommended to buyers.

Among brands that do get recommended, the rates are still low

Even for the 31 brands that received at least one recommendation, the rates are modest. The median recommendation rate among this group is 3%. The mean is 4.8%. The highest in the dataset is 20% (Help Scout). No brand in this study approached the recommendation frequency that the dominant players achieve, because those dominant players are in a separate tier that concentrates AI search volume in a way the mid-tier cannot replicate from a standing start.

This means the recommendation landscape has a sharp step function: a handful of brands are recommended constantly, a small group receives occasional recommendations, and the large majority receive none at all.

Why this happens

AI engines are not summarizing every result they find. They are answering a question. When the question is "what should I use," the engine selects the brand that appears most authoritatively as the recommended answer in the sources it retrieves. The sources that drive recommendations are different from the sources that drive mentions.

A brand gets mentioned when it appears anywhere in a retrieved page: in a comparison table, in a list of alternatives, in a footnote. A brand gets recommended when it is listed as the primary recommendation in the most-weighted editorial sources for that query. That distinction lives at the source level, not the brand level. Brands that rank first or second on the pages the AI engines weight most heavily for buyer-intent queries receive the recommendation signal. Brands that appear lower on those same pages, or on pages the AI engines weight less, receive the mention signal at best.

What it means for how you measure visibility

A brand with a 21% mention rate and zero recommendation rate has the same AI "visibility" as a brand with a 21% mention rate and a 10% recommendation rate, if you are only measuring mentions. They are not the same position. One brand is being recommended to buyers. The other is not.

The distinction matters more as AI search share grows. Mention-level visibility contributes to brand familiarity over many impressions. Recommendation-level visibility drives direct action on a single impression. If AI referral traffic is converting at multiples of organic search, as recent data suggests, those conversions are going to the recommended brands, not the mentioned ones.

Tracking both metrics and knowing which sources drive each outcome is the difference between an AI visibility score that looks reassuring and one that actually predicts buyer behavior.

Methodology note

270 B2B software brands measured in June 2026 using live, search-grounded AI endpoints: ChatGPT (OpenAI Responses API with web search), Gemini (with grounding), Perplexity (Sonar), and Claude (with web search). Each brand received 10 buyer-intent prompts for its specific category. Mention scoring used automated word-boundary matching. Recommendation scoring detected language patterns where the brand was named as the answer to the buyer's question, not merely referenced. Branded queries and research-intent queries were excluded.