AI search optimization services
Answer engines quote sources they can parse and trust. We make a site legible to them: one unambiguous entity, markup that matches the page, and sentences a model can lift without distorting them.
What we actually do
One entity, said once
The same name, description and contact details across the site, the structured data and the profiles that matter, so a model resolves you as one thing rather than three half-things.
Markup that matches reality
Service, product, FAQ and organization data that describes what is visible on the page. Markup that promises what the page does not deliver is worse than none.
Retrievable pages
Answer-shaped sections with a definite statement, a number or a date, and a heading that says what the section is about. Written for a reader first; machines follow.
Measurement
What is indexed, what is retrieved, which pages earn citations, and which claims on your site are unverifiable. Reported per page, not as a score.
What we will not do
- Promise that ChatGPT, Gemini or AI Overviews will recommend you — nobody controls that, and anyone who says otherwise is guessing with your money
- Sell mentions, "AI listings" or paid placements inside model answers
- Publish unverifiable claims to look quotable — that is how a brand ends up cited as an example of a bad source
Usually bundled with technical SEO work, because the same fixes carry both. Priced per site after a short review.
The honest state of this field
AI search is early: the tooling is thin, the measurement is indirect, and results move slowly. That is why the work is worth doing now rather than in a year: the entities that get established first are the ones later described as the obvious answer. We prefer clients who want the foundation built properly over clients who want a graph pointing upward.
Usually looked at together
Technical SEO
Crawl, index and rendering work: the foundation underneath visibility.
Capabilities
The stack we build with and the artifacts you receive.
Selected work
What the work looks like when it is described honestly.
Common questions
- Can you guarantee we get cited by an AI assistant? No, and nobody can. What we can do is make the facts about you present, consistent and readable, then measure what the engines actually return for your buyers' questions.
- How do you measure it? A tracked prompt set: the questions your buyers type, re-run on a schedule across the answer engines, recording who gets named. It turns a vague worry into a number you can watch move.
- What do you actually change? Entity details that agree everywhere, structured data that matches the visible page, and pages written as self-contained answers. We do not write about topics we know nothing about.
Markup is not the mechanism
A 1,885-page study (Ahrefs, published 2026) tracked pages that added JSON-LD between August 2025 and March 2026 against 4,000 matched control pages: no measurable citation lift on any platform (Google AI Overviews, AI Mode or ChatGPT).
Structured data is still worth having: cited pages carry it far more often than uncited ones, and it removes ambiguity about who you are. But it is eligibility, not persuasion. The pages that get quoted answer the question in the first two sentences, in the buyer's own words, with a number or a date in them.
So what we actually spend the time on
- Pages written as self-contained answers, not marketing prose
- The exact questions your buyers type, re-run on a schedule
- One set of facts about you, identical on every surface we can reach
- Markup that matches the visible page, because a mismatch is a liability
Sources: Ahrefs (schema and AI citations, 2026) · schema.org sameAs for entity verification.