Which buyer questions your competitors get cited for, and you do not.

How it runs

1. 01

   You send the domain and three competitors

   Takes a minute. Reply to the email.

2. 02

   We agree the 20 questions with you in writing

   The single most common way a report like this is wasted is asking the wrong questions. So you see the list and change it before anything runs.

3. 03

   We run the scan

   All 7 engines, same day, same wording, your domain and all three competitors scored against the same answers.

4. 04

   We look at the result before we charge you

   If there is nothing useful in it for your category, we say so and you pay nothing.

5. 05

   We write the two pages that need a person

   What kind of pages are winning, and what to do in the next 30 days. Then every URL in the report gets clicked before it goes out.

6. 06

   You get a link within 48 hours

   A page you can forward, not a PDF attachment. Correctable after sending, unlike a file already on your VP's laptop.

What we will not put in it

Every one of these is standard in reports like this and we have evidence against them or no way to measure them.

• A 0 to 100 visibility score

  Computed from 20 questions it moves 5 points when one answer changes. We show a fraction with the denominator visible instead.

• A structured data checklist

  We tested schema, list counts and heading counts on our own corpus. They did not separate cited pages from uncited ones once page length was controlled.

• Traffic or lost revenue estimates

  We do not have your conversion rate and neither do the engines. Every version of that number is invented.

• A percentage improvement promise

  Nobody can honestly make one. Several vendors in this category do.

Questions

What exactly do I get?

A page you can share. 20 is one scraper batch, which is why it is 20 and not more: every one of those questions reaches all 7 engines, and past twenty the extra ones would reach Google alone. You get who is cited, which engines skip you entirely, the exact URLs winning each question you lose, and a short list of what to do about it. Two of those pages are written by hand against your data.

Which engines do you actually ask?

Google AI Overviews, ChatGPT, Perplexity, Gemini, Google AI Mode, Microsoft Copilot, Claude (API). Claude is asked through Anthropic's API with its official web search, which is not the same thing as the claude.ai consumer product, and the report says so.

Why 20 questions and not more?

Our scraper takes 20 prompts in one batch. Past that, the extra questions reach Google AI Overviews and nothing else, so a report selling 50 questions across seven engines would be selling something it cannot deliver. We would rather sell the number that is true.

Do you tell me why I am not cited?

No, and be careful with anyone who does. We tested the usual explanations on our own corpus and they did not hold: schema markup, list counts and heading counts did not separate cited pages from uncited ones once we controlled for page length. What we show is what the cited pages have that yours does not, which is an observation rather than a diagnosis.

Is it automated?

The scan is. The report is not. Two sections are written by a person against your actual results, and every URL in it is clicked before it is sent. That is why it takes up to 48 hours and why we cap it at ten at a time.

What if the report turns out to be useless for my category?

We run the scan before we charge you. If the engines cite nobody in your category, or every page winning is a surface you cannot realistically get onto, we tell you that and you pay nothing. It costs us about 42 cents to find out, which is a fair price for the right to say no.

How do I know it worked?

We store the exact question set, date and engines, so the same scan can be run again. A re-measure is $49 and shows what moved. We do not claim a fix causes a citation, so the re-run is how you find out rather than a promise we make up front.