The pipeline, end to end
1. Query map
We map 20–50 money queries with you: comparison, category, alternative, pain-point and trust queries — not generic brand awareness. These are the queries where “not being in the answer” costs you a deal.
2. Measurement harness
Queries run through ChatGPT, Perplexity, Gemini and Claude on a schedule. Every response is captured verbatim. Built on the open-source measurement stack (open-seo and its companions) — we own the running, judging and diffing.
3. Scoring
Each answer scored on citation presence, position, sentiment and staleness — with the model’s reasoning captured, not just a green/red flag. An LLM-judge pass plus human review; judgment is the product.
4. Diff & fix
Every month we diff against the previous run, then act: structured data, content architecture, llms.txt, third-party citation building. Each fix is tracked against the next run so you can see whether it worked.
5. The one-page report
Citation share on your money queries, month over month — plus what changed, what we did, and what we’re doing next. Written for a non-technical owner. If a number moved, you get the reason. If it didn’t, you get the reason for that.
What the monthly number means
Your headline is citation share: across your money queries and the four AI surfaces, how often you are cited, mentioned, or absent — versus the competitors the answers name instead. It is a defensible number: reproducible from the query map and the raw captures, both of which you retain.
It is not AI referral traffic. We publish the honest picture of AI referral traffic (it is growing fast off a small base) and we will not pretend we can promise it.
What we do not do
- No guaranteed rankings, no “we’ll 10x your AI traffic” — the data doesn’t support it and neither do we.
- No dashboards as a product — dashboards are how this industry offloads work onto clients.
- No black-box “optimization” — every fix is visible and reversible.