How many hours a month an AI visibility client really takes (a task-by-task breakdown)
Almost every agency sets the price of its AI visibility service before knowing how many hours it takes. And it's understandable: the service is new, there's no internal track record, and the pressure to answer the client asking "do I show up in ChatGPT?" pushes you to put a figure on it and get going.
The problem arrives in the third month, when someone on the team does the maths and discovers that the 500 €/month client is eating up 25 hours. This article is that maths, done before signing: the six tasks that make up a month of AI visibility service -what the industry calls GEO (Generative Engine Optimization): measuring and improving how a brand appears in the answers of AI search engines like ChatGPT, Gemini, Perplexity or Claude-, with honest hour ranges, done by hand and with a tool.
The ranges assume a typical client: an SME with a monthly service, 20 prompts monitored across 4 AIs, weekly measurement. If your setup is different, scale proportionally.
Task 1: collecting the data (the one that eats the month)
By hand this means opening each AI, running each prompt, reading the answer, noting whether the client appears, in what position, what's said about them, who else shows up, and saving the screenshot. At about 2 minutes per query, 20 prompts × 4 AIs is 80 queries and around 2 hours 40 minutes a week. Over a month, about 11 hours per client just collecting data. The full multiplication -and why it breaks down with more than three or four clients- is in how to monitor all your clients without screenshots.
With a tool, this task disappears: the weekly query to the 4 AIs runs itself and you receive the structured data. By hand: 10-11 h. With a tool: 0 h.
Task 2: reviewing and interpreting the results
This is the task the client really pays for and the one you must not cut: looking at what changed this week, spotting the new false statement, the competitor that has entered, the prompt where the client appeared for the first time, and deciding what to do with each thing.
By hand it costs more because the data sits in screenshots and scattered cells: comparing with the previous week is archaeology. With structured data and history, it's reading a panel and thinking. By hand: 2-3 h. With a tool: 1.5-2.5 h.
Task 3: adjusting the prompts
Once a month it's time to review the list: has the client launched a new service? Has a competitor appeared that warrants a comparative prompt? Is there a prompt that in three months hasn't yielded useful information? Discipline matters more than hours — changing too much breaks the history, and choosing the list well from the start reduces this maintenance to a minimum. In both cases: 0.5-1 h.
Task 4: building the report
By hand, the monthly report is layout: selecting screenshots, building the progression in a spreadsheet, writing the document. Two or three hours that add no analysis, only formatting. With a tool that generates the report as a PDF, the work shrinks to what does count: reviewing, adding your interpretation and the next steps. Exactly what that document should contain is in the anatomy of a good AI visibility report. By hand: 2-3 h. With a tool: 0.5-1 h.
Task 5: the meeting (the one you must not cut)
Thirty or forty-five minutes of meeting plus its preparation. This line item is the same with or without a tool, and rightly so: the meeting is where the client perceives the value, and cutting it to save costs is saving in the exact spot where renewal is decided. In both cases: 1-1.5 h.
Task 6: content and corrections
The work that moves the needle: correcting false data in the sources, updating listings and directories, securing mentions, and publishing content the AIs can cite. By hand, writing 2-4 decent monthly pieces plus the corrections easily comes to 6-10 hours. With a tool that generates the content drafts, the human work is left at reviewing, tailoring to the client and publishing, plus the source corrections. By hand: 6-10 h. With a tool: 2-4 h.
The total, and what it does to your margin
| Task | By hand | With a tool |
|---|---|---|
| Data collection (weekly) | 10-11 h | 0 h |
| Review and interpretation | 2-3 h | 1.5-2.5 h |
| Adjusting prompts | 0.5-1 h | 0.5-1 h |
| Monthly report | 2-3 h | 0.5-1 h |
| Meeting and preparation | 1-1.5 h | 1-1.5 h |
| Content and corrections | 6-10 h | 2-4 h |
| Total per client per month | 22-29 h | 5.5-10 h |
Now the margin. At an internal cost of 25 €/hour, the by-hand client costs between 550 and 725 €/month in hours. If you charge 500-600 €/month -a usual range for SMEs, the full numbers are here-, you're paying to work. With a tool, the hours cost 140-250 € and you have to add the licence: in Surfeo for agencies, 20 €/month for the base account plus 35 €/month per client (40 prompts, 3 AIs, 6 articles a month and weekly tracking) or 79 €/month (75 prompts, all 4 AIs, 16 articles). Total per client: 175-330 €. The service goes from loss-making to having a margin that justifies selling it.
Notice which hours the tool removes and which it doesn't: collection and layout disappear -the ones the client neither sees nor values- and interpretation, the meeting and editorial judgement remain — the ones that justify your invoice. That's the proof the maths is right: if an "optimisation" takes hours of analysis or meetings away from you, you're not optimising, you're degrading the service.
Frequently asked questions
Does the first month take the same?
No: add the initial audit (3-5 hours by hand) and the setup — defining prompts, baseline, plan. Reckon on the first month taking between 150% and 200% of a normal month, and plan it that way in what you deliver in the first month.
Can I go below 5-6 hours a month with a tool?
You can: by dropping the meeting, reducing interpretation to forwarding the PDF without reading it, and not touching the sources. But then the client is paying for something they could hire directly, and the renewal is on borrowed time. The 5-6 hours are the floor of a service that deserves to be called a service.
How many clients can one person handle?
By hand, at 22-29 hours per client, one full-time person handles 2-3 clients and not much more. With a tool, between 8 and 12 depending on the intensity of the service — meaning a single person can run a portfolio billing several thousand euros a month.
Do I charge by the hour or as a fixed package?
A fixed package, almost always. The value of the service isn't in the hours but in the outcome of the work, and charging by the hour penalises you precisely when you become efficient. Define fixed deliverables (weekly measurement, monthly report, X content pieces, corrections) and put a monthly price on them.
Before pricing your next AI visibility service, do this maths with your real numbers. And if you want to see how much of the "by hand" column you can save, start with the pricing page and run the figures against your current portfolio.
