Last updated: 28 August 2026
Most conversations about AI visibility start with a promise. This one starts with a baseline — measured, unflattering, and public.
Key takeaways
- Across 700 live AI engine scans of four UK brands' target keywords (July–August 2026), the brands were named in just 3% of answers.
- Brand maturity dominates: an established brand with 600+ published articles was mentioned in 6% of scans; two brands in their first week scored 0%.
- The engine read the brands' websites far more often than it named the brands — being cited as a source comes weeks or months before being recommended by name.
- Average quality score across 863 scored, published articles in the same period: 81.2/100 — publishing quality does not shortcut the authority timeline.
What we measured
Our platform runs scheduled citation scans: it puts a brand's real target keywords to a live AI engine and records whether the answer names the brand, what it cites, and who else appears. This dataset covers 700 scans across four UK brands — a security services company, a marketing agency, a B2B software product and an AI software product — between July and August 2026.
Honesty notes, because our methodology requires them: these scans ran on one engine (Claude with live web search) during this window, as API access to other engines was constrained; a further 83 rows in our database from demo environments were excluded from every figure here. Every number states what was counted out of what.
The baseline: 3% of answers name the brand
Of 700 scans, the target brand was named in the answer 23 times — 3.3%.
If that number feels low, it should. These are not vanity queries — they are the buying-intent keywords each brand actually targets, which means competitive answers where the engine has many established names to choose from. This is the honest starting line for a growing brand in 2026, and any provider quoting dramatically higher day-one figures should be asked to show their scan data.
Maturity is the strongest signal we can see
| Brand profile | Scans | Named in answer |
|---|---|---|
| Security services — 600+ articles published, 15+ years trading | 270 | 6.3% |
| Marketing agency — 800+ articles, newer brand entity | 270 | 2.2% |
| B2B software — first week of publishing | 80 | 0% |
| AI software — first week of publishing | 80 | 0% |
The pattern is consistent with everything else we measure: engines learn sources before they learn brands. A new brand's articles can be read, used and even cited by an engine while the brand itself goes unnamed. The gap between "the AI read my website" and "the AI recommended me" is where the real work of GEO lives — entity building, third-party evidence, and time.
What doesn't shortcut the timeline
Publishing quality, on its own, does not. The 863 scored articles published across these brands in the period averaged 81.2/100 on our quality gate — structurally sound, sourced, machine-readable content — and the baseline is still 3%. Quality is necessary (unreadable or unverifiable content never gets cited at all), but authority compounds on a calendar, not a sprint.
What this means if you're starting now
- Set baseline expectations honestly. If a vendor promises immediate AI recommendations, ask for their scan data.
- Measure from day one. A 0% baseline is not failure — it's the number your progress gets judged against. Without it, nobody can prove anything worked.
- Watch citations before mentions. Your site appearing as a source in AI answers is the leading indicator that naming comes next.
- Invest where engines look for proof — reviews, communities, press, public records — not only on your own pages.
We'll republish this baseline as the dataset grows and engine coverage widens. The methodology, as always, is public — and if you want to know where your own brand stands today, the free audit runs this exact kind of scan on your market.