Generative Engine Optimization
Being ranked is no longer the same as being recommended
When an assistant answers a buying question it names two or three vendors and moves on. There is no page two. We measure whether you are in that answer, find out what it quoted instead, and change the pages until you are.
What an engagement looks like
Five stages. The first two exist so that everything after them can be argued with data rather than opinion.
1. Baseline the prompts your buyers actually type
Not keywords — questions. We assemble the prompt set from your sales calls, support tickets and the queries already converting in Search Console, then run them across every major assistant and record the answers verbatim.
2. Measure share of answer
For each prompt: are you mentioned, in what position, with what sentiment, and which competitors appear beside you. That is the number the work moves, and it is measurable before we change anything.
3. Trace the citations
Assistants quote sources. We record which URLs they pull from in your category — your pages, competitor comparison articles, review sites, forums — because that list tells you where the influence actually lives.
4. Restructure to be quotable
Answer-first sections, passages that survive being lifted out of context, explicit entity statements, verifiable claims, and schema that agrees with the copy. Most of this also helps classic rankings, which is why it is rarely wasted spend.
5. Fix access and keep score
Crawler access for GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot and Google-Extended reviewed deliberately, an llms.txt where it earns its place, then monthly re-runs against the same prompt set.
What we report
- Share of answers that mention you, by assistant and by prompt
- Sentiment and position of the mention inside the answer
- Which of your URLs get cited, and which competitor sources replace them
- AI-assistant referral sessions in analytics
Questions about AI search
Is GEO just SEO with a new name?
It overlaps heavily and diverges in one important way. Classic SEO optimises a document to rank. GEO optimises a passage to be retrieved and quoted inside a generated answer, where there is no position one and no click to measure. The underlying content quality work is shared; the measurement and the page structure are not.
How do you measure something with no rank tracker?
By running a fixed prompt set across the assistants on a schedule and scoring each answer: mentioned or not, position within the answer, sentiment, and the URLs cited. Same prompts, same cadence, so month two is comparable to month one.
Should we block AI crawlers?
That is a business decision, not a technical one. Blocking protects content from training use and also removes you from the answers. What we insist on is that the choice is deliberate — most sites we audit are blocking or allowing by accident, via a robots.txt nobody has read in two years.
Does this replace our existing SEO work?
No. Assistants ground their answers largely in pages that already rank and in third-party sources that already have authority. Weak classic SEO is usually the reason a brand is invisible in AI answers too.
Find out what the assistants say about you
We will run your top prompts across ChatGPT, Gemini, Perplexity and AI Overviews and send you the answers verbatim, with the citations.