In our own September run of 60 clinic queries, every single one returned an AI summary. Only 33% of the recommendation-type queries named a clinic at all — the rest passed over them without a name. Which side you are on is a measurement, not a guess.
9,000
AI responses sampled per brand each month across three engines, daily. Not one screenshot.
What we don't claim
No one has shown that appearing in AI answers causes more customers. Our data can't show it either, and anyone who says otherwise is selling.
Here is what we do know. The cost of entering a channel while the slots are still empty is not the cost of entering once they are full. Measuring now leaves you a baseline — the only way to answer "what did it look like before" when it starts to matter.
Every tracked query runs repeatedly across engines, so every number arrives with its sample size.
We extract every source the engines cite — so you see which sites feed AI answers in your category, and where you're absent.
Where your figure sits in the distribution for the same city and vertical. A position in a distribution, not a judgment of quality.
04 — PRESCRIBE
One thing to fix first
We compare where you're absent in the citation map against the conditions our own measurements show actually produce a naming. You get page specs and drafts — one priority a month, not a backlog.
05 — RE-MEASURE
We tell you when it moves
After the work, we re-measure against the baseline: up, unchanged, or down. After that, we tell you when a number moves outside its confidence interval.
We write the findings; the provider approves what gets published. We don't run your channels, and what you pay has nothing to do with which way the number goes.
The question this answers
The report your agency sends is written by your agency. When the party running the channels is also the party scoring them, that isn't verification. We write findings but never run the channels — and when we check whether our own findings moved the number, we hold back untouched control queries.
Baseline preserved → work happens (you or your agency) → same panel, re-measured → month-over-month Δ
5% of the retainer verifies the retainer.
For each city × vertical we collect every source the AI answers actually cited and publish it by domain. It shows which blogs, communities and platforms feed the answers in a category — and which of them you are missing from.
Query counts, schedule, engines and scoring rules are documented. Anyone can re-run our measurement and get the same answer. The prescriptions come from the same published rules — and when we score whether our own advice moved a number, we keep untouched control queries alongside it.
Query text, timestamps, full responses, every citation URL — exported as CSV or JSON for every run. A number you can't re-check isn't evidence.
We don't control what the engines output. Anyone guaranteeing AI visibility is selling what measurement cannot honestly promise. What you pay has nothing to do with which way the number goes.
Your clients are already asking what the AI-search line item is doing. Hand them an independent verdict under your own brand.
Verdict PDFs and a client portal under your logo. The independence stays ours; the relationship stays yours.
One flat fee covers ten client accounts — no per-seat math when you win a new client.
Open a pitch with the prospect's real AI-search coordinates — measurement first, promises second.
Every plan starts with a free snapshot. Cancel monthly — the data you exported stays yours.
· English & Japanese panels (multilingual coordinates)
We ship fixes as tools you own, but never run your channels, sell retainers, or charge on results. We partner with agencies — we don't compete with them.
Rank trackers watch links on a results page. We sample the generated answers themselves — whether you're mentioned, how, and which sources the engine cited.
Every query, timestamp, engine, full response and citation URL — for every run, as CSV or JSON.
English, Korean and Japanese today. Others on request.
Is there a contract term?
No. Cancel monthly. The responses, citations and time series you've exported stay yours after you leave.
Free snapshot: one brand, 25 queries, a 4-page report. No credit card.
Run a free snapshot