People now ask an assistant instead of a search box, and it answers with three sources. Being one of those three is a different problem from ranking, and it is measurable.
- 01
- AI citations
- 02
- Retrievability
- 03
- Entity clarity
- 04
- Claim evidence
- 05
- Monitoring
- R/01
Retrievable beats persuasive
A retrieval system pulls passages, not pages. Content that is self-contained per section — a claim, its evidence and its context in the same few sentences — gets retrieved. Content whose meaning depends on four paragraphs of build-up does not.
We restructure key pages into passages that stand alone, which also happens to make them easier for humans to skim.
- R/02
Claims a model can verify
Assistants favour statements that are specific, dated and attributable. 'Industry-leading performance' is unusable. 'Reduced median frame time from 250ms to 66ms on Intel Iris Xe, August 2026' is a citable fact.
We go through your claims and either attach evidence or cut them.
- R/03
Knowing what the assistants currently say
You cannot improve what you have not measured. We run a fixed prompt set across ChatGPT, Perplexity, Gemini and Google's AI Overviews, and record whether you are mentioned, how you are described and which competitors appear beside you.
That baseline repeats monthly, so the effect of the work is visible rather than asserted.
Deliverables
06 items
- 01
Citation baseline
A fixed prompt set run across the major assistants, recording mentions, descriptions and competitors.
- 02
Passage restructuring
Key pages rewritten into self-contained sections that survive being retrieved out of context.
- 03
Claim evidence pass
Every marketing claim either given a specific, dated, attributable source or removed.
- 04
Entity graph
Consistent organisation, product and person data across your site and the sources assistants read.
- 05
Monthly citation report
Movement in mention rate and share of voice per assistant, against the same prompt set.
- 06
Misdescription fixes
When an assistant describes your business wrongly, we trace the source and correct it.
Queries
Retrievable content, verifiable claims and entity clarity, so assistants cite you rather than a competitor.
The underlying content quality work overlaps heavily with good SEO. What is genuinely new is the measurement — you can now ask the assistants directly what they say about you, repeatedly, and track it. That baseline is the part most companies do not have.
No. You can influence what it retrieves by making accurate, well-structured, verifiable content the easiest thing to find, and you can correct upstream sources that are wrong. Anyone promising control is describing something that does not exist.
They share most of the same work. AEO targets extracted answers inside a search results page; GEO targets citation inside a generated response. We usually run them together and report them separately.
Next
Send us the domain and the queries you want to win. We will come back with what is actually blocking you, whether or not you hire us.
AI citations · Retrievability · Entity clarity · Claim evidence · Monitoring