What to deliver in the first month of an AI visibility service
The first month is the most dangerous one in an AI visibility service. The client has just signed, pays from day one, and the results — showing up in answers from ChatGPT, Gemini, Perplexity or Claude — take months and on top of that nobody can guarantee them. If the client sees nothing tangible that first month, the seed of cancellation is planted before the work has had a chance to bear fruit.
The solution isn't to promise faster. It's to deliver visible work every week, making clear from the proposal stage that the first month delivers work, not appearances. In GEO (Generative Engine Optimisation, the discipline of improving how a brand shows up in answers from AI search engines) this is even truer than in SEO: AI answers are volatile and they move on sources that take time to update.
Here's what we deliver, week by week.
Week 1: the baseline
Deliverable: the initial audit, dated.
The whole service will be measured against this snapshot, so it gets done properly or it's worthless: 15-25 prompts covering the sector's commercial questions, run across all 4 AIs, with systematic notes on whether the client appears, what's said about them, which data is wrong, which competitors come up and which sources each AI cites. You've got the full procedure, with checklist, in the first audit step by step.
What the client receives on the Friday of week 1: a 6-10 page document with the executive summary, the annotated screenshots and the coverage table (how many prompts they appear in, by AI). If the result is the usual one — in our study of 9,865 Spanish SMEs, 91% only appeared in 1 of the 4 AIs (full study) — the document already justifies the service on its own.
Why this deliverable matters more than it looks: without a dated baseline, three months from now you won't be able to prove improvement. The baseline is your renewal insurance.
Week 2: the source diagnosis and the gap list
Deliverable: the gap report — why they don't appear and where to act.
The AIs build their answers from sources: the client's website, their listings and directories, their reviews, third-party mentions. Week 2 audits that ecosystem and cross-references it with the sources the AIs cited in the audit. The result is a prioritised list, along these lines:
- Incorrect data in circulation (the most urgent: every week it goes uncorrected, someone gets the false data).
- Inconsistencies between sources — the website says one thing, the Google listing another, a directory a third.
- Absences from the sources the AIs cite for their sector: directories, comparison sites, media.
- Hard-to-extract website content: pages that don't clearly answer what the company does, where and for whom, with no structured data.
- A shortage of reviews and third-party mentions compared with the competitors who do appear.
What the client receives: the list, in plain language, with each gap explained in a sentence and its priority. This turns the "you don't appear" of week 1 into a "you don't appear because of this", which is far more reassuring: problems with a cause have a solution.
Week 3: the 90-day plan and the first fixes
Double deliverable: the approved plan + the quick fixes already done.
The 90-day plan arranges the gaps into a calendar: what gets fixed in month 1, what content gets published in months 2-3, what gets measured each week. Important: the plan commits to actions, not results — which KPIs you can put in writing and which you only report we cover in which KPIs to put in a GEO proposal.
And alongside the plan, the first technical fixes already done, because some work doesn't need to wait for any creative sign-off:
- Correcting the false data found in the sources you control (website, Google listing, main directories).
- Unifying name, address, opening hours and services across all sources.
- Adding structured data to the website's key pages.
- Rewriting the home or services page if it doesn't answer "what does this company do, where and for whom" in an extractable way.
What the client sees: a concrete before/after ("the AI said you closed on Mondays; the source of the error was directory X; it's now fixed"). It's the first moment in the service when something changes, rather than just being measured.
Week 4: the first monthly report and the second measurement
Deliverable: the report that sets the rhythm for the months ahead.
The full battery of prompts is run again — same list, same 4 AIs — and compared with the baseline. The monthly report has four blocks:
- What's been done (audit, gaps, fixes, plan): the month's verifiable work list.
- What the AIs say today vs week 1: coverage by AI, changes detected, corrected false data that's already reflected (or isn't yet — and why that's normal).
- Market context, so the client understands the playing field: AI Overviews have cut organic CTR by an average of 61% in Spain and traffic from AI has grown 527% in a year (ismajimenez.com). The channel moves; that's why it gets measured every week.
- Next month: what comes next from the 90-day plan.
The expectation to set at the month's closing meeting: in month 1 there are almost never visible changes in the answers — the AIs take time to pick up the fixes. Saying it before the client asks is what separates a serious service from one that improvises.
The data-gathering trap (and how to avoid it)
Notice that weeks 1 and 4 depend on running dozens of prompts across 4 AIs and noting everything down in a comparable way. Done by hand for one client, that's hours; for the whole roster, every week, it's a multiplication that doesn't add up. We built Surfeo for agencies so that part runs automatically: each client in their own space, their prompts run every week against the AIs (40 prompts and 3 AIs on the €35/month per-client tier; 75 prompts and all 4 AIs on the €79 one), with comparable history and a PDF report ready for your reporting — and the weeks of the first month stay focused on the work the client values: diagnosis, fixes and plan.
Frequently asked questions
What if the client demands to see results in the answers in the first month?
That conversation belongs in the proposal, not in week 4. The first month's commitment is the four deliverables in this article; changes in the answers typically arrive between months 2 and 4, and irregularly, because they depend on when the AIs pick up the corrected sources. If you promised appearances in 30 days during the sale, the problem isn't with the deliverables plan.
How many agency hours does this first month take?
With automated measurement, between 10 and 15 hours: 4-5 for the audit and the document, 3-4 for the gap diagnosis, 3-4 for fixes and plan, 2 for the monthly report. Without automating, add another 6-10 hours just running and noting prompts. Against a fee of €400-600/month (the ranges, here), the maths works out as long as you're not paying for the gathering hours yourself.
Does the first month include creating new content?
Usually not: month 1 is about measuring, diagnosing and fixing what already exists. Citable content — pages that answer the sector's questions, material that third-party sources can pick up — comes in months 2-3 of the plan. Bringing it forward without the diagnosis done is writing blind.
What happens if after the first month the client wants to stop?
They walk away with the baseline, the diagnosis and the fixes done — real work worth what they paid for it. But the structure of this first month exists precisely so that doesn't happen: a client who has received four dated deliverables in four weeks has reasons to stick around for month 2.
Got your first meeting coming up and need the baseline now? Drop the client's website into the free AI visibility test and you'll have the first snapshot in minutes — the first deliverable of the first month, even before signing.
