Brands are named in AI answers 2.7x more often than they are linked
Across 90,000 answers, brands were named in the answer text 2.7x more often than they appeared as a linked citation. Equivalently, link-counting captures only about a third of that named-brand presence, and the gap is widest on ChatGPT and narrowest on Perplexity.
- Method
- 9,000 prompts × 10 samples across six engines, entity extraction on answer bodies
- Sample
- 90,000 answers
The measurement gap
Most AI visibility tools count citations: the links in the source list or chip carousel. But an AI answer can recommend a brand by name, in a sentence, without linking it anywhere. To a buyer reading the answer, that mention is the visibility. To a link-counting tool, it does not exist.
Method
Nine thousand prompts, ten samples each, six engines. For each answer we extracted linked citations and separately ran entity extraction over the answer body against a 4,000-brand dictionary with disambiguation on category context.
The gap, per engine
| Engine | Mention-to-link ratio | Answers naming a brand with no link to it |
|---|---|---|
| ChatGPT | 3.6x | 71% |
| Google AI Mode | 3.0x | 64% |
| Microsoft Copilot | 2.6x | 59% |
| Google AI Overviews | 2.3x | 55% |
| Baidu ERNIE | 2.2x | 53% |
| Perplexity | 1.4x | 27% |
Why it matters commercially
A brand that is consistently named but rarely linked has a different problem from a brand that is neither. The first has been absorbed into the model's category knowledge and needs its description corrected; the second is absent and needs to exist somewhere the engines read.
Those are opposite workstreams, and a link-only tool cannot tell them apart.
What we return
Sentiment is the obvious next question
We are extending this dataset with claim-level extraction — not just whether a brand is named, but what the answer asserts about it. Early signal suggests the variance in how brands are described is larger than the variance in whether they appear at all.
Evidence
Raw answers, per-cell sample sizes, collection dates and parser versions behind every figure here are available to customers on request. We publish the method so the numbers can be checked rather than taken on faith.
See this for your brand
Every figure here comes from the same measurement behind our dashboard and API. Tell us what you want measured and we'll show you where your brand stands.