Method
How we check, step by step
The method is deliberately boring and repeatable. It has to be. Your score this month only means something if next month's score was produced in exactly the same way.
The four steps
You can check every one of them
We read your site and write the questions
Buyer questions, not brand questions — what someone asks when they are deciding what to buy and have not heard of you yet. You see the whole set before we ask a single one, and you cut the ones your buyers would never ask. A question you could not plausibly win only makes your own score look worse than it is.
We ask the chatbots, with web search turned on
Each question is asked the way a buyer would ask it. We save the full answer and every website the chatbot read before answering. We keep the raw answers, so any number in your report can be traced back to the words that produced it.
We count what was said
Every company named, the order they were named in, and every page used. Then a checking step compares each number in the report against the saved answers. If a number in the text disagrees with the record behind it, the report will not print.
You get the report, and on Growth, the work
How often you were named, who was named instead, which pages decided it, and what is open. On Growth the report also carries the month's work already written: six publish-ready pages and four outreach emails, drafted for your approval.
The question set
Locked, so the months are comparable
Once you approve the questions we freeze them and record a fingerprint of the set. The same questions are asked the same way every month. If a question is ever added or changed, the report says so on the front page and the affected comparisons are marked.
This is the difference between a measurement and a demo. A tool that quietly rewords its questions can show you any trend you like.
| What is fixed | Why it matters |
|---|---|
| The questions | Reworded questions produce different answers, and the change would look like progress. |
| How many times we ask | The same question can get slightly different answers. Asking more often gives a steadier number. |
| Which chatbots | Each one reads the web differently. Adding one halfway through the year breaks the comparison. |
| The counting rules | Named is named. A mention in passing is not the same as a recommendation, and the report separates them. |
The report
What arrives in your inbox
Your score
How many of the questions named you, where you appeared in the answer, and what moved since last month.
Who was named instead
The full leaderboard for your questions, and whether the leaders are competitors, marketplaces or guides.
The pages that decided it
Every source the chatbot read, listed at the back, so you can open them and check the numbers yourself.
What is open
The questions that named nobody at all — usually the cheapest ground to take, and the ones we write for first.
Plain dealing
What nobody else will tell you
- Nobody can control what an AI says. What can be changed is which websites it finds. Anyone promising more than that is selling something else.
- One report is a snapshot. Answers shift each time you ask. The change over months is the real signal, which is why we lock the question set and never edit it between runs.
- We reach the chatbots the way software does, not by typing into the app like your customer. The answers are very close, but not identical, and we say so in every report.
- Answers depend on where the buyer is sitting. A question asked from the United States can miss companies a British buyer would be shown. Every report says which country we asked from.
- The first check is free. Not a trial with a card on file. If it tells you nothing useful, that is the end of it and you owe nothing.
See this for your own business
Send us your website address. We will write the questions your buyers ask, ask them, and send you the whole report — not a teaser.