What this does
Your twenty most important queries get run through the AI answer engines on a schedule. The workflow records who gets cited, captures the answer text, and diffs it against last week.
You get a short report: gained, lost, unchanged. Not a dashboard to interpret — a change list.
Why a diff and not a score
AI answers are non-deterministic. A single run is an anecdote. One query returning you once is not a measurement, and a “visibility score” built on one sample per query is a number that will mislead you.
This runs each query multiple times and reports the mention rate with its sample size — cited 3/5 — then compares rates week over week. That is the difference between a signal and a
horoscope.
What you get
- The workflow file and the query list template
- The sampling configuration, with a note on how many runs each engine needs before the rate stabilises
- A setup document covering API keys, rate limits, and what to do when an engine changes its response format
- A history retention note: keep the raw answers, because the interesting signal is often in how you were described, not whether
Honest limits
Engines change their retrieval behaviour without warning, and a retrieval update can move your citation rate overnight for reasons that have nothing to do with your work. The workflow tells you that it changed. Working out why is the consulting part.