AI answer tracking as an agency deliverable
The client question has already arrived: “our CEO asked if we show up in ChatGPT — can you check?” Agencies that answer with a screenshot look reactive; agencies that answer with a trend line look like a platform.
The service in one sentence
A scheduled prompt set per client, results stored per market and engine, rendered as a monthly report: mention rate, citation share, who got recommended instead.
What the report needs — all in the response
result.text— the verbatim answer to quote: “when users ask X, ChatGPT says…”entities[]— the client and competitors namedsources[]— the pages cited, which becomes the content action listsearchQueries[]— the searches behind the answer, which doubles as an SEO briefcountry/usState— per-market splits for multi-region clients
Nothing in that list requires infrastructure you maintain. A batch call submits the month’s prompts; webhooks deliver results; your report template does the rest.
Pricing the deliverable
Usage-based credits make the marginal cost of an extra client a few dollars a month. A typical agency shape: 50 prompts × weekly × one market per client — a few thousand credits per client monthly, inside even the mid-tier plans across a book of clients. Charge it as a monitoring retainer or fold it into the SEO scope.
The pitch artifact
The fastest client acquisition tool in this category: run a prospect’s 10 category prompts before the pitch meeting and bring the answers. “ChatGPT recommends these three competitors when asked about your category” closes scopes better than any deck. See the white-label setup for the platform version, and mentions vs citations for the metrics to put on the first slide.