A monthly GEO report has five parts: a one-line headline on where the brand stands, its mention and recommendation rates beside two or three competitors, the sources the answers cited and who owns them, what you shipped this month, and what happens next. Say what the numbers are, keep the panel constant, and do not promise causes.
Key takeaways
- Lead with a rate and a comparison. A brand's mention rate means little until it sits beside two or three named competitors.
- Show the sources. They explain the number and they show the client which of your actions matter.
- List what you shipped. The report is where the work becomes visible, and a rate can stay still in a month with real progress.
- Keep the question set constant between months, and change it only on purpose, saying so.
- State the method in two lines: the date, the number of questions, the engines. A number without them will not survive a challenge.
The point of a monthly report is that it answers the client's actual question. For GEO the question is short: "does AI recommend us yet?"
A good report answers that in the first line, shows the evidence, and makes it plain what you did to earn your fee. Here is a structure that does it in a page or two.
1. The headline
One sentence with a rate and a direction. "In April, your brand was named in 14 of 40 buyer questions across four engines, up from 11 in March."
Say it in the client's terms. They do not care about the acronym. They care whether a buyer who asks an AI for a supplier hears their name.
2. The comparison
A rate on its own is hard to read. 35% might be good or bad. Put the client beside two to four named competitors on the same questions, for the same period.
This is the table clients forward internally, because it turns "how are we doing" into "how are we doing against them".
Show two rates. Mention: the brand was named. Recommendation: the brand was put forward as a choice. The gap between them is informative. Named often but recommended rarely means the engine knows the brand and has no reason to prefer it.
3. The sources
This is the section a plain rank report cannot give, and it is where you show your thinking.
List the pages the answers cited most, and mark who controls each: the client, a competitor, or an independent publisher. If the answers lean on review sites and lists the client is absent from, say so. It tells the client why publishing more on their own site is not the whole answer, and it justifies work off their site. How to see which sources an AI answer is built from covers how to read them.
4. What you shipped
Pages published or rewritten. Facts added to existing pages. Third-party pages you pitched, and any that replied. Technical fixes, such as a crawler that was blocked.
Rates can stay flat through a month of real work, because answers take time to shift and because they vary. The shipped list is what keeps the report honest in a flat month and stops the client reading a flat number as a flat effort.
5. What happens next
Three or four concrete items for next month, each with a reason that points back to the sources section. Not "continue optimization".
Two lines on method
Put a short note at the foot: the date the answers were collected, how many questions, which engines. Then add a sentence saying answers vary between runs and that the report is directional. Why the score moves between runs explains the evidence for that, and it is the sentence that saves you when a client says "I asked it myself and got a different answer".
Keep the question set the same from month to month. If you add questions, say so, and show the old set separately, or the comparison stops meaning anything.
What to leave out
Rank position in an AI answer. It is the least stable number available.
Traffic you cannot attribute. Google says clicks from its AI features are counted in the Web search type in Search Console, mixed with ordinary search, so do not present them as AI-only.
Causal claims. "Our new page got you cited" is a claim you can rarely support. "We published this page in March, and the June run cites it" is a fact you can.
Doing it faster
Assembling this by hand for one client takes an afternoon. For six it takes a week you do not bill. GeoShark produces the report as a shareable link or a PDF from each run, with the sources, the comparison and the drafted fixes in it, and you add the two lines of judgement that only you can. There is a sample report to look at, and the client report page describes what is in one.
For the wider argument on what to sell, see GEO vs SEO: what changes for a small agency. If you are new to all this, start with what GeoShark and GEO are.
FAQ
- What should a GEO report for a client include?
- A headline on where the brand stands, its mention and recommendation rates against two or three named competitors, the sources the answers cited and who controls them, what you did this month, and what you will do next. Add a two-line method note giving the date, question count and engines.
- How often should I send an AI visibility report?
- Monthly suits most retainers. Collect more often than that, weekly is common, so a monthly report reflects several runs and one unusual day cannot swing it. Send a shorter note in between only if something changes that the client should hear from you first.
- Can I show AI visibility in Google Search Console?
- Only partly. Google says sites appearing in AI features are included in the overall search traffic in Search Console, in the Performance report under the Web search type. That means the clicks are there, but mixed in with ordinary search, so it cannot give you the AI share by itself.
- Should the report promise that the fixes will get the brand recommended?
- No. Show what was done and what the next run measured. The research on generative engine optimization reports gains from citing sources, quotations and statistics on a benchmark, not in a client's category, and answers vary run to run. Let the client see the movement rather than promising it.
Sources
- AIs are highly inconsistent when recommending brands or products — SparkToro (Rand Fishkin), 2026
- AI features and your website — Google Search Central, 2026
- GEO: Generative Engine Optimization (Aggarwal et al.) — arXiv, 2024
- client reporting
- GEO
- agencies