Monthly · cancel at 30 days
Move one number, every month.
Your named rate — how many of your 40 buyer questions produce an answer containing your brand. Same panel, same engines, every month. If it stops moving, you cancel.
- Monthlyper market · no minimum term
- 30 daysnotice to cancel
- Day 90stop-billing trigger
- 2–4pieces shipped a month
The monthly loop
Measure, ship, corroborate, report.
- 01Re-run the panelSame 40 questions, same 5 engines, same day of the month.
- 02Ship on-site2–4 pieces, weighted to the questions you are losing.
- 03Work the corpusCorrect the record where retailers and marketplaces have it wrong.
- 04Report one numberNamed rate by engine, deltas inside the noise floor marked as noise.
The shape of a month under retainer. No engagement has run a full cycle yet — when one has, the results page is where it shows up.
Every month, against the same baseline
The stop-billing clause — what happens if the number does not move
- Day 0–90You pay. We measure against your noise floor.
- Day 90Not improved beyond the floor? Invoicing stops.
- Day 90–180We keep working. You pay nothing.
- Day 180Third month refunded either way. Still flat, and you should leave.
Trigger is aggregate across all five engines, and it goes in the engagement agreement — not only on this page.
Boundaries
What we do not do.
Stated up front, because the fastest way to waste your money is to buy the wrong thing.
No volume content
A thin page gives a model nothing to lift, and a large body of unquotable content is work that is not doing any.
No prompt injection tricks
We do not use hidden text, instruction-shaped copy or manufactured reviews. They are detectable, and the downside when they are caught lands on your brand, not ours.
No unreviewed claims
Nothing about SPF, concentration, weight limits or age ranges ships without your compliance sign-off. An engine repeating a wrong number of ours becomes your problem.
Before you commit
Why a retainer instead of paying per result?
A corpus accrues monthly and decays if abandoned, so we carry the risk instead of you. Cancellable at 30 days, no minimum term, and a written stop-billing clause: if your named rate has not improved beyond the noise floor after 90 days, we stop invoicing, keep working free for up to 90 more, then refund the third month. The trigger is aggregate across all five engines and it goes in the engagement agreement.
Does this replace our SEO agency?
No, and you should not let it. Rankings have not stopped driving discovery, and the two share technical foundations. We coordinate rather than compete — the panel data tells your SEO vendor which of their pages models actually read.
How much content do I have to review?
Two to four pieces a month, none of it shipped without your approval. Regulated claims need sign-off from your compliance owner — we will not publish a specification you have not confirmed.
What does the monthly reporting actually look like?
One page. Named rate by engine, this month against last, with deltas inside the noise floor marked as noise. Then quotes, who was named instead, factual errors, what we shipped, and assistant referral traffic. Built to be forwarded without editing.
What if a model update moves the number instead of your work?
It will, both directions, and nobody can fully separate the two. We record the model version on every run, re-run within a week of a major release, and report against a noise floor. Where a jump is plainly a model update, the report says so — an agency claiming every rise as its own is telling you it is not measuring carefully.
How is this different from an AI visibility tracking tool?
A tool tells you the score. It will not correct your specification on Amazon.co.jp, argue a distributor into re-uploading it, or tell you which of the five causes is yours. If you have a team to act on a score, buy the tool. Buy us to move the number, not report it.
Do you work with our existing agencies?
Yes, and we would expect to. Ask each of them whether they measure which sources engines read in your category; if one does, that is useful to know. We hand them the panel data either way. It is your data.
Find out what AI says about your brand.
Send us your brand and category. We run 20 buyer questions across five engines by hand and send you the transcript within 48 hours.
- What arrives
- 20 questions, 5 engines, every answer quoted in full
- Who sends it
- Snow Lee, by hand, from Seoul
- Your email
- Used to send the report and reply about it. Privacy.
| Buyer question | Engine | Named |
|---|---|---|
| best Korean sunscreen for oily skin | ChatGPT | not named |
| K-beauty sunscreen that does not pill | Claude | named |
| Korean skincare brands sold at Ulta | AI Overviews | not named |
| …17 more questions, with the verbatim answer and every source cited | ||