Our Method — How We Engineer AI Citations
A four-step system. Repeatable per client. One client per vertical, per market.
Our method has four steps: audit your inclusion across every major AI engine, fix the technical foundation (SSR, schema, entity, llms.txt), engineer the off-site corroboration that earns named citations, then measure continuously so the position compounds instead of decaying. We run it for one client per vertical, per market.
- Step 01
Audit — see what the engines see
We run a live, multi-engine baseline: ChatGPT, ChatGPT Search, Perplexity, Gemini, Claude, and Google AI Mode. We log inclusion rate, cited share, and the named-source frequency for the prompts your customers actually ask. The output is a numbered report — not a vibe.
- Step 02
Foundation — fix the floor
SSR, robots posture, schema.org coverage, llms.txt, a single canonical entity (Person + Organization with stable @id), and extractable 40–60 word answer blocks on every page that matters. Without this, nothing else compounds.
- Step 03
Citation engineering — earn the named source
We build the off-site corroboration the engines look for: third-party mentions, structured datasets, expert bylines, and a citation graph that mirrors how real authority is signalled. This is the part that turns retrievability into being the answer.
- Step 04
Continuous measurement — defend the position
Inclusion is not a rank you set and forget. We re-run the prompt set monthly across every engine, track slope, and densify wherever a competitor is gaining cited share. The compounding moat is built in this step.