Generative Engine Optimization
When somebody asks ChatGPT or Perplexity who to hire, it comes back with three or four names. We work out why it picks those and get you into the set.
Being findable is not the same as being recommended
A buyer who used to type a query into Google now asks an assistant to shortlist for them. The assistant comes back with three or four names and a sentence on each. That shortlist is assembled from what the model has read about you across the web, not from where you rank, and most companies have never once checked what it says.
The uncomfortable part is that you can rank first for every keyword you care about and still be absent from the shortlist. Models lean on what third parties say about you, on whether your category and your claims are described consistently wherever they appear, and on whether your own pages state plainly what you sell and who for. Fixing it is reputation work and structural work at the same time.
You cannot be picked from a list you are not on.
Six things we do
01
Prompt testingWe run the prompts your buyers use across ChatGPT, Gemini, Perplexity and Claude, and record who gets named and in what terms.
02
Citation auditWhich third-party pages the models draw on for your category, where competitors turn up and you do not, and which of those we can realistically influence.
03
Source placementGetting you accurately represented on the directories, comparisons and roundups the models keep quoting back.
04
Claim consistencyThe same description of what you do, who for, and with what proof, everywhere a model might read it. Contradictions are what get you dropped.
05
On-site evidencePages that state your category, your clients and your results in language a model can lift without having to guess at it.
06
Re-testing on a cycleModel answers drift, and they change without telling anyone. We re-run the same prompts monthly and report the movement, including when it goes against us.
3 of 4
Assistants naming the client after one programme
Progress is counted in mentions, not positions
There is no ranking report for this. The honest measure is how often the assistants name you for the prompts that matter, which is why we baseline those prompts before starting and re-run exactly the same ones afterwards.
Answer Engine Optimization
The same structural groundwork, aimed at the one direct answer Google gives rather than what an assistant recommends.
Learn more →AI SEO
The ranking and technical layer underneath. Models read the pages that crawlers can reach.
Learn more →Website Development
If a model cannot render or parse your pages, the evidence sitting on them does not count for anything.
Learn more →Find out what the assistants say about you today.
The free audit runs your own prompts across the major assistants and shows you the answers, alongside your rankings and where your paid spend is going.