AI visibility tracking for agencies: a weekly GEO brief for every client

Why are clients asking about AI visibility now?

Because their buyers ask an AI assistant for a shortlist before they search, and a business left out of that answer never sees the lost click in analytics. Clients notice when a competitor gets named and they do not. They want someone to measure it and tell them what to fix, and that someone is usually their agency.

• The question arrives as "are we showing up in ChatGPT?", which a single spot check cannot answer honestly.
• Answers vary between runs and between engines, so the only credible answer is a rate measured the same way every week.
• The follow-up work, pages, listings and comparisons, is work agencies already sell.

What does the tracker measure for each client?

For each client it measures mention rate per engine, share of voice against the competitors you name, rank by first mention, citation rate, and the domains cited in answers that left the client out. It stores every raw answer and link, so you can recount after fixing an alias and quote any wrong fact exactly.

• 10 to 30 buyer questions per client, each with a permanent id so the trend line stays comparable.
• Three samples per prompt, per engine, per week by default, so each number is a rate rather than one lucky answer.
• A noise test: a change is called a trend only when it is larger than normal run-to-run variation.
• The measurement method is written up in full in the guide to measuring AI search visibility.

• How to Measure AI Search Visibility: A Repeatable Method

How do you run one tracker per client?

Create an environment for the client, deploy the AI visibility tracker template into it, and fill in that client's business, aliases, competitors and prompts. Each environment keeps its own secrets and keys, so no client's answers, history or credentials mix with another's, and you run every client from the same console.

• The template shows the call count and an estimated cost before the first run, and asks you to approve that budget.
• A per-run cap keeps each collection inside the budget you approved for that client.
• Collection runs early on Monday and resumes by itself if a run is interrupted, so the brief is ready when you start the week.
• Once a month the agent checks whether an engine's model has been retired or replaced, and proposes the change rather than making it silently.

• AI visibility tracker template

What does the client actually receive?

The client receives what you send them. The agent delivers its Monday brief to you, in Slack or Telegram. You read it, add your own recommendations, and forward it under your name. Clients do not get a login or a dashboard: the deliverable is the brief and your judgment on top of it.

• Mention rate per engine, with last week's rate beside it.
• The prompts the client gained or lost this week.
• Any factual error an engine made about the client, quoted exactly.
• The domains cited in answers that left the client out, and up to three recommendations tied to the evidence.

How should an agency price AI visibility tracking?

Price it like any retainer: add up what one client costs you each month, then divide by one minus your target margin. The platform cost is the client environment's hosting plus the engine calls, which scale with prompts times engines times samples. Twenty prompts, five engines and three samples is 300 calls a week.

• Your time is the larger cost: reading the brief, writing the recommendations and doing the fixes.
• Weekly runs keep the engine cost modest; daily runs multiply it by seven for little extra signal.
• Sell the fixing, not only the measuring: the brief names the pages and listings worth your client's money.
• The agency pricing calculator turns your cost and margin into a setup fee and a monthly retainer.

• Pricing calculator

What can't the tracker do for a client?

It measures what the engines say through their APIs with web search on. That is a consistent panel for trends and sources, not a copy of any one person's screen, since the consumer apps add memory and personalization. It also does not change what the engines say. The fixing is still content and listings work.

• It is not a client reporting portal: there is no client login and no white-labelled dashboard.
• It cannot guarantee a mention. No one can, and a brief that promises one should not be forwarded.
• For a client whose buyers never ask an AI assistant, classic search tracking is the better use of their money.

Which Qoren plan does an agency need for this?

Ultimate ($169/mo, up to 10 environments) and Business ($499/mo, no environment limit) are the plans built for running several clients, each in its own isolated environment, and both include API access. One tracker per client means one environment per client, so the environment limit is the number to check against your roster. Every client's hosting and engine calls draw on one account budget, with alerts or an automatic pause when it runs out.

• Qoren pricing
• AI Automation Agency Platform: Run Agents for Clients