Generative Engine Optimisation
Your buyer asksan AI now.Are you in the answer?
One list comes back, and it is short. We ask ChatGPT, Gemini, Perplexity and Claude the questions your buyers actually type, then count how often your name appears — and whose appears instead.
Free. One field: your domain. No email needed to see your score.
How it works
You give us a domain. We do the rest.
Four steps, about three minutes. The one rule we never break: your brand name is not in the question. Try both below.
- SmileCare London
- Chiswick Dental Studio
- Bupa Dental Care
We read your site
Sector, market and services come from your own pages. No forms to fill in.
sector dental · market West London · services implants, invisalign, hygiene
We write the questions
Real purchase-intent questions across the funnel — and your brand name never appears in them.
q23 how much do clear aligners cost in w4 · decision
We ask the engines
Live search on, identical prompt and temperature across every engine, three separate runs each.
perplexity run 2/3 · 4 brands · 9 sources
We count
How often you are named, in what position, with what variance — and who is named instead.
named instead: SmileCare London, Chiswick Dental Studio
What makes this different
Most AI visibility tools ask once and estimate the rest.
We read two competitor reports line by line before building this. Three things they get wrong — and what each one looks like when it is done properly.
We ask three times, not once
Ask the same question tomorrow and a different set of brands comes back. A tool running one query per engine cannot see that, so it labels you “inconsistent” and moves on. We run every prompt three times and publish the variance band.
A single run would have reported one of these three as the truth.
Real citation URLs
Competitors publish an estimated source breakdown and admit it in a footnote. We collect the actual URLs and group them by who owns them.
What the internet says about you
We open the third-party pages the engines cite, pull out the checkable claims, and compare them with your own site. A wrong opening time in a directory is a small thing until it is the thing an engine repeats.
A short report when there is little to say
If you are almost never named, the competitor matrix and share-of-voice tables would be noise. We cut them and the 90-day plan focuses on earning your first appearance. Twelve useful pages beat fifty padded ones.
One full audit
Every answer stored raw, so the score can be recomputed without asking again.
Pricing
Two depths. Same engine.
The free audit tells you your score. The full audit tells you what to do about it.
Free audit
8 questions × 3 engines × 1 run
- GEO score and maturity stage
- Which engines name you and which do not
- Your top three competitors in AI answers
- Your three biggest visibility gaps
- A shareable result page
50 questions × 4 engines × 3 runs = 600 answers
- The full prompt library — every question with funnel stage, visibility rate, variance band and gap type
- Citation source breakdown from the real URLs the engines returned
- Narrative integrity — third-party claims checked against your own site, with the fix for each contradiction
- Technical audit: schema, E-E-A-T, AI crawler access, internal linking
- Competitor matrix and relative share of voice
- A 90-day plan in three sprints, each with a score target
- PDF you can hand to a client or a board
You buy the full audit from your free result page.
Questions
Things people ask before they run it
No, and we will not pretend otherwise. We query the engines through their APIs with web search enabled. API answers are close to, but not identical with, what a logged-in user sees in the consumer app — personalisation, memory and app-only features all differ. Every report says this in its methodology section. Most tools in this category share the limitation and do not mention it.