MMethodology
How we measure.
AI visibility is full of big claims and screenshots. Ours come with receipts. This is the discipline behind every report we send — published, so you can hold us to it.
- best building management company in Mayfair—named
- who handles out-of-hours cover in London—named
- what should a facilities contract includenamednamed
- how much does managed security cost——
- recommend a building manager for a listed buildingnamednamed
- who is liable for compliance in a managed block——
The questions never change. So the difference is the finding.
AI answers drift week to week, so a single screenshot proves nothing. We build a fixed panel of real buyer questions for your market, run it on a schedule, and report trend lines. The panel stays fixed so every re-run is comparable — and you can re-run it yourself.
Mention rate, citation rate, share of voice — each metric states what was counted out of what, on which engines, with run dates. If a stat can’t be verified, it doesn’t go in the report.
Before work starts we agree what "working" looks like — specific numbers, specific dates — and record them. At each checkpoint we publish results against those criteria, whether they flatter us or not.
Every client account carries a registry of claims that must never publish: unverified statistics, invented quotes, superlatives without evidence. The engine blocks them at generation and again at publish time.
Every article is scored across structure, evidence, machine-readability, linking and readability. Nothing publishes below 80. The scoring is automated, so it never gets tired or generous.
Engine access changes — APIs open and close. Your reporting always states exactly which engines were queried and when, and what could not be measured. A disclosed gap beats a confident guess.