ChatGPT, Perplexity, Claude and Gemini do not return ten blue links. They answer, and they quote a handful of sources while doing it. Being one of those sources is mostly SEO, plus a specific layer on top. We build the layer, and we measure whether you are actually getting quoted.
Four engines, one prompt set, measured monthly.
and
AI engines draw disproportionately from sources that already rank organically. So most of what earns a citation is the work that earns a ranking: a crawlable site, real topical depth, content that answers the question properly, and links you actually earned. In our own retainers that is roughly four fifths of the effort. The remaining fifth is specific to AI search, and it is the part almost nobody is doing yet.
Which is why we always recommend both: SEO as the foundation, and this on top of it. Booked on its own, the additional fifth is structure and schema layered over nothing.
Every item here serves rankings and citations at the same time. None of it is AI-specific.
This is the additional scope on this page. Smaller than the foundation, and the reason you get quoted rather than skipped.
None of this replaces your SEO plan. It changes what the same content has to look like, and it adds the measurement nobody else is running.
Written to be quoted, not just to rank.
A model quoting you has to be able to lift a sentence and have it stand on its own. That means explicit definitions, comparisons with named alternatives, and numbers that carry their own context. We rework existing pages for this and brief new ones this way from the start.
These engines read about you, not only from you.
AI answers pull from review sites, industry press, comparison pages and forums as much as from your own domain. So part of the work sits off your site: getting named, correctly and consistently, in the places these models sample.
Cheap groundwork, done properly once.
llms.txt, schema markup, server-rendered content, and no critical text hidden behind client-side rendering. None of this is heavy lifting, and none of it works retroactively, so we do it early and then stop talking about it.
The part almost nobody does.
We build a prompt set from the questions your buyers actually ask, run it against ChatGPT, Perplexity, Claude and Gemini on a fixed schedule, and log whether you are cited, in what position, and who is cited instead. That log is the whole basis for deciding what to do next.
Clearly defined phases, concrete deliverables, transparent cost.
We build the prompt set for your category and run it. You get a documented starting point: which questions get answered, which sources the engines quote, and where you appear or do not. Most companies have never seen this number.
We sort the gaps into two piles: things your SEO roadmap already fixes, and things it does not. Only the second pile becomes additional work. We would rather scope this small and honest than invent a separate project.
Restructuring the pages that matter, shipping llms.txt and schema, and working the off-site mentions. This runs alongside the SEO retainer, not instead of it, because the same content has to earn a ranking and survive being quoted.
The prompt set runs again and the AI Visibility Score goes into your reporting next to rankings and traffic. What moves gets extended, what stalls gets reworked or dropped. Citations shift with model updates, so this is a standing measurement, not a one-off audit.
Not running SEO yet? Start with SEO and add this on top once the foundation holds.
Reproducible measurement, transparent methods, raw answers you can read yourself. No voodoo, no black boxes.
Our own tracking scripts run your prompt set against ChatGPT, Perplexity, Claude and Gemini on a schedule and store every answer, so the score is reproducible and you can read the raw responses.
Schema validators for structured data, Screaming Frog for render and crawl checks, Google Search Console for the organic side of the same foundation.
30 minutes. We run a short prompt set for your category before the call and tell you who the engines are quoting instead of you. You decide afterwards whether it fits.
Free growth analysisNo newsletter, no tracking spam.