26 October 2026
What the report is for
The client wants three answers: does AI recommend me, who does it recommend instead and why, and is the work making a difference. Every section below answers one of them. Anything that doesn't can go in an appendix.
The template, section by section
1. Summary
The score from 0 to 100, the margin of error, and the change since last month in one sentence that says whether it's meaningful: “From 29 to 22, within the ±12 margin: no meaningful change.” Then the one or two things that matter most this month.
2. Score per assistant
ChatGPT, Perplexity, Gemini and Claude separately. They read different sources, so a low score on one assistant points to a specific problem: Perplexity not finding the website, for example.
3. Competitors and why AI picks them
The businesses AI recommends, scored in the same answers, and the reasons it gives: opening hours, reviews, prices on the website, a specialty. This section is the to-do list, because it shows the facts AI found about them and not about your client.
4. What AI knows about the business
The answers to the questions with the client's name: does AI know it, and are the facts right? A wrong phone number or old opening hours goes at the top of the actions.
5. Sources AI consults
Split in three: listings and directories where the client can be, forums and reviews, and other businesses' websites. The first group is where to act this month.
6. Website readiness
A short checklist: AI crawlers allowed, content readable without JavaScript, business markup (schema.org) with address, phone and hours, an llms.txt file, the city and the contact details as text.
7. Actions
What was done since the last report, what's next, and in what order of impact. The next measurement confirms what the website now has, so the client sees work turning into checks.
8. Evolution
The score month by month, next to the most recommended competitor, with the margin drawn as a band. A line inside the band hasn't moved; a line that leaves it has.
Appendix: the questions and the answers
Every question asked, and each assistant's answer. Few clients read it all, but having it there makes the rest of the report credible.
Five rules that keep the report honest
- The same questions every month. Change them and the months stop being comparable. Extra questions the client cares about can be shown, but outside the score.
- Always the margin of error. It protects you in a bad month and makes a good month believable.
- One page of summary. The detail is there for whoever wants it.
- In the client's language, and with the questions their customers ask in their city.
- No positions promised. The report shows what was measured and what was done.
Without building it by hand
The Chatlisted report has these sections, filled in with the client's data every month or every 15 days. On the Agency plan, it carries your agency's brand and goes out as PDF too. How to check whether ChatGPT mentions a brand explains how the score and the margin are worked out.