If your agency just sold a GEO or AEO retainer, you now own a delivery pipeline. That is the part nobody walks you through at the point of sale. This post is for small agencies selling AI-visibility retainers to their clients: what the pipeline actually costs to staff, what a white-label ops layer covers, and the mechanics of who owns what.

The problem you have

The market data is public and the numbers are high. First Page Sage’s GEO & AEO pricing report (published October 7, 2026; the publisher is itself an agency — take the figures as the best market snapshot available, with that conflict of interest noted) compiled 1,146 price quotes from 214 agencies (https://firstpagesage.com/seo-blog/geo-aeo-pricing-statistics/). Its Starter-tier median retainer is $2,950/month — reporting, a technical audit, and a few pages a month, with a typical three-month minimum term.

You priced yours around or above that number, probably correctly. The uncomfortable part: the quote you sold includes a measurement pipeline that did not come with the contract. Someone has to map the client’s money queries, run them on a schedule across the AI surfaces (ChatGPT, Perplexity, Gemini, Claude — one tool’s coverage is not “the market”), capture and score the answers, diff against last month, decide which findings matter, implement the fixes, and turn it into a client report the account owner can send without a translation pass. On a $2,950/month retainer with three-month minimums, that is a recurring delivery line item on every client you book — and it does not get cheaper at scale the way a tool license does.

The staffing math

Two sides of the same ledger.

Side one: staffing it in-house. The same First Page Sage report models in-house delivery from its survey of in-house GEO/AEO program leaders: an in-house single GEO/AEO specialist costs $146,028 per year — a $96,400 median salary with 27% benefits and payroll taxes, worked out by the report’s authors. Output on that model: 3.1 deliverables per month, and 13.6 weeks to the first one, most of it recruiting. (Same source, same agency-publisher caveat.)

Run that against a realistic small-agency book: at ~$12,000/month all-in, one specialist is the whole pipeline — every client, query map, run, diff, fix, and report. That works margin-wise only if one person serves several $2,950+/month retainer clients, has never taken PTO, and is not the person your other services depend on. And the 13.6-week recruiting clock starts before the first client deliverable ships.

Side two: the tool cost underneath. The actual measurement stack — query runs, captures, diffs — runs on open-source tooling at roughly $10–20 per client per month in hosting and API usage, on our numbers from running it. A commercial AI-visibility tool adds a few hundred dollars per month per seat if you want your team in a polished product; the honest cost comparison is in our tools-vs-managed-service post.

The gap between side one and side two is judgment work: choosing the client’s money queries, running them monthly, deciding what a finding means for that client’s revenue, implementing the fixes, and writing the report. That is a service, and it is the part a retainer quote implies.

What the ops layer includes

The white-label offer is that judgment work, run on clients you own, reported under your brand:

  • Query map per client. 20–50 money queries — comparison, category, alternatives, pricing, trust — mapped with you or the account owner. Not a keyword list; the queries where “not being in the answer” costs the client a deal.
  • Scheduled runs on four surfaces. Every query run across ChatGPT, Perplexity, Gemini and Claude on a monthly schedule; every response captured verbatim and retained.
  • Scoring and diffs. Each answer scored on citation presence, position, and sentiment, with the model’s reasoning captured; the whole map diffed against the prior month.
  • Fixes implemented. The actual work — structured data, comparison and pricing content, docs, llms.txt — scoped, executed, and verified against the next run.
  • The client report under your brand. The one-page format: citation share with the MoM delta, what changed, top three fixes, what was done. Your name, your client, your margins.

The stack is open and cheap — that is stated on the about page and is the whole point. You are not buying a black box; you are buying someone running it, making the calls, and owning the number while you run the agency.

Is this the right fit?

It fits if: you have a handful of GEO/AEO retainer clients (two to ten is the natural band), no dedicated GEO/AEO staffer, and clients in or near the B2B space whose money queries fit a 20–50 query map. The report format and the four-surface measurement are identical for every client, which is exactly what makes it run as an ops layer rather than a bespoke project per account.

It does not fit if: your engagements promise enterprise-scale deliverables (multi-market, multilingual, digital PR volume at the top of the market’s tier structure — that is a staffed team, not an ops layer), or if the strategic point of the retainer is building in-house capability at the client. In the second case, the honest advice is to staff it or partner on the staffing, because outsourcing the learning is not what they bought.

The margin check before you sign anything: the per-client ops fee has to leave your agency a real margin at your retainer price on a per-client basis — not on average across the book. Run that arithmetic on one client before discussing the rest.

Mechanics

  • Who owns the client: you. Always. Clients buy from you, pay you, and stay yours. We are invisible in the relationship; nothing in the reporting names us unless you say so.
  • Who owns the data: you and the client. The raw captures, scorecards and query maps are delivered to you with every report. Nothing is held hostage — if we part ways, the measurement trail stays with the engagement.
  • Reporting cadence: monthly, tied to the run schedule. When a number moves, you get the reason in writing before the client asks. When it doesn’t move, you get the reason for that, and the next experiment.
  • The contract: a simple per-client monthly fee between your agency and us, scoped per client. No volume commitments to start — one or two clients is a normal starting point. Setup (query map + harness stood up + first scorecard) is a one-time line item per client.

One structural note: we are based in India, working async. The fit is a small agency with a handful of retainer clients, not a house with two hundred.

Start a conversation

If you have one or more GEO/AEO retainer clients and the delivery pipeline is the open question, send the basics: client count, rough retainer size, and the query landscape per client. We will tell you straight if the fit is wrong.

CTA: hello@aecited.com — or read how the measurement actually runs on the about page.