METHOD
AI engines choose sources two ways: they extract passages that directly answer the question, and they synthesize from sources the wider web treats as authoritative. We work both. On-site, we restructure your pages into extractable answers. Off-site, we build the citations, mentions and evidence that make engines treat you as a source.
We'll run your market's prompts live on the call.
01
We collect the questions your buyers actually ask an AI — from your sales calls, search data and market language — and fix them into a tracked set of 20–50 prompts. The set is frozen for the engagement, because frozen prompts are what make month-over-month numbers comparable.
02
The same prompts go to ChatGPT, Gemini, Perplexity, Claude, Google AI Overviews and Meta AI on a schedule. Per-engine logging is non-negotiable: only 11% of domains cited by ChatGPT are also cited by Perplexity (AuthorityTech), so a blended score would hide the movements that matter.
03
Every answer is reduced to two auditable numbers: mention rate and citation share. Two numbers, defined in plain language, trackable month over month — that's the whole instrument, and it's deliberately simple enough to argue with.
How often your business is named in the answers to your tracked prompts — named, not necessarily linked.
Your share of the cited sources behind those answers, measured against named competitors.
04
You get one page of numbers: both metrics per engine, the competitor comparison, which prompts moved and which didn't. The report below is a sample with demonstration data — every client report has this exact shape, with real figures.
MONTHLY VISIBILITY REPORT
SAMPLE — DEMONSTRATION DATA34%
Mention rate
12%
Citation share
| MONTHLY VISIBILITY REPORT | Mention rate | Citation share |
|---|---|---|
| ChatGPT | 40% | 14% |
| Gemini | 33% | 12% |
| Perplexity | 45% | 18% |
| Claude | 27% | 9% |
| Google AI Overviews | 38% | 11% |
| Meta AI | 20% | 8% |
05
Each report ends in a prioritized fix list, and the next month's work is that list — restructuring, schema, crawler access, off-site placements. Then the next run measures whether it moved anything. That loop is the entire service.
Answer engine optimization restructures your own pages into extractable answers — direct openings, citable evidence, schema, crawler access. It's in your control, and first movement typically shows in four to eight weeks on retrieval-based surfaces.
The on-site work →Generative engine optimization builds your citation footprint across the sources engines synthesize from — reviews, references, communities, coverage. It takes three to six months, and it's the defensible part: evidence-dense, citable material is what the Princeton GEO research found engines reward (Aggarwal et al., 22–41% visibility lift).
The off-site work →We build a fixed set of buyer-intent prompts for your market, run it on a schedule across ChatGPT, Gemini, Perplexity, Claude, Google AI Overviews and Meta AI, and log two numbers per engine: mention rate, how often you're named, and citation share, your share of cited sources versus competitors. You see both trend monthly.
Both metrics are defined in plain language in the glossary, and the monthly report format is shown — with clearly labeled demonstration data — on the how it works page.
First movement typically shows in four to eight weeks on retrieval-based surfaces like Perplexity, ChatGPT search and Google AI Overviews, because on-site answer fixes register on the next crawl. Authority-driven visibility, where engines recommend you from synthesis, builds over three to six months. We report both monthly, so you watch the trajectory rather than wait on a promise.
No, and you should walk away from anyone who says yes. Engines change source selection without notice, and only 11% of domains cited by ChatGPT are also cited by Perplexity — nobody controls that. What we guarantee: a defined prompt set, measured monthly, the scoped work delivered, and honest reporting of what moved and what didn't.
The 11% overlap figure is from AuthorityTech's 2026 per-engine citation audit. It's also why we track six engines separately rather than selling a single blended score.
Book a 30-minute call; bring one question your customers would ask an AI.