Research · 5 min
When AI Did Not Name These Brands, Half the Time It Named a Rival
Of 39 brands no AI engine named, 21 had a competitor named in their place and 18 had nobody named at all. The second group is the more interesting one.
Everything we know about getting recommended by AI, in one place.
Guides for understanding the measurement, articles on what moves the numbers and a glossary that says what each word on the report means.
Start here
Guides
How a scan works
25 buying questions, 5 engines, one deterministic score. Ten minutes to understand the whole method.
Docs: every number, defined
Engines, questions, KPIs, versions, tiers, monitoring, data. The reference page.
What we measure and what we won't do
The limits of the method, the rules we won't break to sell you something and what happens when we get it wrong.
We audit our own brand first
The founder teardown: our own scan, warts included.
Read a full report before you run one
Real scans of public brands, every section unlocked.
From the blog
Articles
Methodology · 5 min
What Our Score Is Made Of, Including the Factor We Often Cannot Measure
Twelve weighted dimensions, the exact numbers, and the re-normalisation that stops an unmeasured dimension counting as a zero. Read off lib/scoring.ts, not a marketing page.
Research · 6 min
We Scanned 55 Brands Across Five AI Engines. Thirty-Nine Were Named by None of Them.
Every brand that ran a LensCore scan in September 2026, measured across five AI engines. 70.9% were named by zero engines. The script that produces these numbers ships with the post.
For your kind of business
What the audit checks for you
Glossary
The words on the report, defined
- AI visibility
- Whether an AI engine names your brand when a buyer asks it a question in your category. Not the same as ranking on Google.
- Buying question
- A question someone types into an AI engine when they are ready to choose: “best running shoes for flat feet”, “which CRM for a 10-person team”. LensCore asks 25 of them per scan.
- GEO
- Generative Engine Optimisation: the work of getting AI engines to recommend you. The AI-era counterpart of SEO.
- Share of voice (mentions)
- The share of buying questions in which your brand was named at all.
- Recommendation rate
- The share of questions in which the engine put you first or second, not merely mentioned you.
- Prominence
- Where in the answer you appear. First is worth more than fifth; the report weights it.
- Citation
- When an engine links to a source for its answer. Citation rate is how often your own pages are that source.
- Entity
- The single, unambiguous idea of your brand that engines corroborate across your site, Wikidata, schema and third-party mentions. A weak entity is the commonest reason a brand is not named.
- Live search
- Whether the engine actually searched the web for its answer or replied from memory. Every answer in a LensCore report records which it was.
- Lenscore Retrieval Engine
- Our own fifth engine: a search-and-synthesis pass over an independent web index, so the four frontier engines are not the only witnesses.
- Deterministic score
- Same inputs, same score. The score moves when you or the engines change, never because of mood or randomness on our side.
- Methodology version
- A stamp on every report saying which measurement rules produced it, so two reports are only compared like for like.
Reading is the slow way to find out.
Free scan · 90 seconds · no credit card required.
Run a free scan