How it works
We put twelve of your buyers' real questions to Claude with live web search. Then we fix what AI assistants read about you, and ask the same twelve again.
The heart of it
Measure · Week 1–2
A number is not the work. It is the brief for it.
The baseline gives you three things: the score, the list of questions you lose, and the businesses named where you are not.
Get you in
Then we get you in, question by question.
We give the assistants a reason to name you, working down your gap list one question at a time.
Prove it
Then show you the number.
With monitoring, the same measurement runs every month, giving you your score, your curve and every question set beside last month's. It is written in plain language, and none of it asks you to take our word for it.
The first 90 days
The first ninety days, week by week.
Week 1–2
The baseline
We build your question panel and run the first measurement. You end the fortnight with a score, the list of questions you lose, and what we would do about each.
Week 3–6
The foundation
We clear up the contradictions, make your business legible to the assistants, and write the first answers to the lost questions worth the most to you.
Week 7–12
The traces
Your answers and mentions start appearing in the places the assistants read from. Once your fixes are live, the same panel is asked again: the first reading that can move.
Day 90
The comparison
The before and after: the baseline against the reading once your fixes are live, in plain numbers, with the list of what comes next. With monitoring, a reading every month from here.
Before you start
What people ask about the engagement.
How long before anything moves?
Four to eight weeks before the panel shows movement, and about three months before a change is large enough to argue with. Assistants pick up new information on a delay, and no amount of spending compresses that. The trade is that it is equally slow to undo: once you are the named answer, displacing you is somebody else's ninety-day project.
What do you actually need from me?
Less than you would expect. Access to publish changes on your website, or someone who can apply them; the facts about your business that should be consistent everywhere; and an hour in the first fortnight to agree the question panel. After that the work, and the monthly reading with monitoring, run without you unless something about the business changes.
What happens in the first month?
We build the question panel, run the first measurement, and hand you the baseline: your score, the list of questions you are absent from, and which businesses get named instead. Nothing is optimized before the baseline is taken, on purpose. A number taken after the work has started is not a baseline, and without a baseline nothing afterwards is provable.
What if the score does not move?
Then the report says the score did not move, and we say why we think so. That is the whole reason the panel is frozen at the start: a measurement you can fail is the only kind worth running. Where the cause is something we got wrong, fixing it is our cost, not a change order.
Am I locked into anything?
No. The check is free, the audit is a one-time fixed price, optimization is a fixed scope you approve before work starts, and monitoring is monthly and cancellable at the end of any month. There is no minimum term on any of it, which is deliberate: the measurement is supposed to tell you whether to keep going.
Can you work alongside our existing agency?
Usually, and often it is the right arrangement. Most of the foundation work — site structure, speed, structured data — is already theirs, and duplicating it wastes your money. What we add is the measurement and the work on quotable answers and third-party mentions. We will tell you which of the fix list they should own.
Step one costs nothing.
In 48 hours the check tells you which customer questions you are missing from. Closing that gap is the work we do next, if you want it.