The AI visibility slide you should add to every report from this month on
There are two ways AI visibility ends up in your reporting meetings. The first: the client brings it up, usually as "my brother-in-law asked ChatGPT and we didn't come up", and you improvise. The second: you bring it, on a slide you prepared, and the client thinks "my agency is ahead of the curve".
The difference between those two scenes is a single slide that takes less than half an hour to put together. This article is that slide, element by element.
Why get ahead of the question (and why this month)
The question is coming; the only doubt is who puts it on the table. Frequent ChatGPT use in Spain has gone from 4% to 28% in two years (Funcas, III AI Survey, 2026) and, according to the INE, 37.9% of the population used generative AI in the last quarter of 2025. Among your client's customers, and among their brothers-in-law, people are already asking AI about their sector.
When the client asks first, you're playing catch-up: now you have to explain why you weren't already looking at it. When the answer is already in your report, that same data point positions you as the agency watching the new ground before anyone asks. Same work, opposite position. If the question has ever caught you without an answer, we have the full script for that meeting; this slide exists so it never happens again.
The slide, element by element
A single slide. Not a section of eight; that comes once the service is signed. Three elements:
1. The metric: presence in X of 4 AIs
At the top, large, a single number: "[Brand] appears in 2 of the 4 main AIs for its 10 key questions".
Why this metric and not another:
- It needs no explanation. "2 of 4" is something any managing director processes in a second. "Share of voice in generative engines" is not.
- It's honest about the volatility. AI answers change from one week to the next; an aggregate presence metric across several questions holds up better than "you came third in such-and-such prompt", which next month may have shifted without meaning anything.
- It's almost never 4 of 4, and that's what opens the conversation. In the study we ran on 9,865 Spanish SMEs, 91% only appeared in 1 of the 4 AIs (full study). Your client, barring a surprise, has visible room to improve right there in the number.
The "10 key questions" are natural-language prompts — "best employment law firm in Murcia", "which company would you recommend for X?" — that you define once and repeat every month, always the same, so the number stays comparable.
2. The screenshot: a real answer, with a date
Below or beside it, a screenshot of a real AI answer, with the date visible. One, not six.
Which one to pick depends on the story that month: if the AI recommends the client, that one (a win to show); if it recommends a competitor and not them, that one (urgency to activate); if it says something wrong — old opening hours, an outdated address — that one without hesitation, because it's the finding that drives action most.
The screenshot does what the number can't: the client sees the AI talking about their market with names and surnames. It's the difference between "your visibility could be better" and "look, whoever asked this on the 4th got sent to your competitor".
3. The line: context plus action
Close the slide with a single line stating what the data means and what gets done about it. Template:
Key data
"Traffic from AI answers converts at 14.2% versus 2.8% for classic organic (sector data collected by roymo.es). Next step: [specific action]."
The specific action varies: "fix the sources giving the old opening hours", "work the reviews the AIs cite in this sector", "extend measurement to the whole service catalogue". The point is that the slide never ends on the problem; it ends on what your agency is going to do about it.
What the slide should not include
- Ten metrics. Mentions, sentiment, average position, share of voice... all of that exists and has its place once the service is running. On the opening slide, one metric. We cover the rest of the reporting hierarchy in what to show in the AI reporting meeting.
- Untranslated jargon. If you write "GEO", define it alongside: optimising your presence in AI answers. Better still: don't write it yet.
- Promises. None of "we'll get you into all 4". Nobody controls AI answers; you work the sources and measure the evolution. The slide promises monitoring and work, not guaranteed results.
Two versions depending on your situation
If you already offer the AI visibility service: the slide evolves. The number becomes a series ("2 of 4 → 3 of 4 in three months"), the screenshot shows a new mention won, and the line reports what was done: corrections applied, content published, sources worked. It's the slide that justifies the retainer month after month.
If you don't offer it yet: the slide is a baseline plus declared monitoring. Number, screenshot and a line along the lines of: "We're monitoring this channel; when the data justifies acting, we'll bring a proposal". At no cost to the client, it positions you as the one who saw the change coming, and leaves the upsell ready for the next quarter: when you propose the service, you'll bring three months of your own data that nobody else has. How to turn that slide into a contract we cover in how to sell AI visibility to your SEO clients.
How much it costs to produce each month
By hand: 10 prompts × 4 AIs × capturing and logging each answer, per client, every month. The first time is manageable; across your whole book it's a full morning nobody pays you for. That's exactly the work we automate with Surfeo for agencies: each client's prompts run on their own every week across the 4 AIs, the history is stored and the PDF report comes out ready to attach to the reporting. The slide goes from half an hour per client to copying one figure and one screenshot.
Frequently asked questions
What if the number is bad and I scare the client?
A bad number, well framed, is the best argument: "you're like 91% of Spanish SMEs, and that's exactly why it's an opportunity, not an emergency". What scares people isn't the data; it's finding out from a relative that their agency wasn't looking at it.
How often do I update the figure on the slide?
Reporting is usually monthly, but the underlying measurement should be weekly: AI answers are volatile and a single monthly reading can catch an odd week. We cover the right rhythm in how often to monitor AI visibility.
Doesn't this force me to know GEO before I sell it?
For this slide, no: you need to measure and report, not optimise. The "I don't offer the service yet" version is precisely how you learn the ground with real data before committing to working it.
What do I do if the client wants "the whole slide" as a service?
Congratulations: that's the sale. Move from the slide to a serious initial audit and a proposal with deliverables and a price. Don't give away the full service inside the SEO report without charging for it, or you'll have created the expectation that it's included.
The slide takes half an hour to prepare and the data is free: run the AI visibility test on your client's site and come away with the number and the screenshot for this month's reporting.
