A weekly AI visibility report for agency clients: setup, cost per client and a CSV you can hand over

Disclosure: I built the Apify Actor in this post. Written with AI assistance.

Clients now ask a new question: "are we showing up in ChatGPT?" An answer from a screenshot is not a report. A weekly, repeatable measurement per client is, and the cost per client is easy to work out in advance.

AI Brand Monitor on Apify is priced per AI answer. Here is a setup you can copy. Full-size runs need a paid Apify plan.

Setup per client (about 5 minutes)

1. Open the Actor page and fill in the client's Your brand domain, Competitor domain(s) (3 or 4) and Queries to track (6 questions their buyers ask).
2. Engines to query: Perplexity, ChatGPT and Gemini. Samples per query: 2. Query language: the client's market.
3. Run it once, then save the input as a task named after the client.
4. In Apify Console open Schedules, click Create schedule, use the cron 0 8 * * 1 (every Monday at 08:00), and add the client's task.
5. In the task, set Max total charge as a hard ceiling.

Repeat for each client. The ready-made weekly monitoring task is a good starting point.

Cost per client

A citation record is one engine x one question x one sample, and it costs $0.05 (less on higher Apify plans: Bronze $0.045, Silver $0.04, Gold and up $0.035).

6 questions x 3 engines x 2 samples = 36 records = $1.80 per client per week, about $7.80 a month. On the Gold rate the same run is $1.26. Five clients at the base rate come to $9.00 a week. If you prefer one run for two clients, the ready-made agency task (2 brands, 15 questions each, 3 engines, 2 samples) is 180 records = $9.00, so $4.50 per brand.

What goes into the report

Each record has what a client wants to see:

• brandCited and competitorsCited: is the client named, and who is named instead.
• citationShare: the client's share of citations against the competitors.
• the source URLs with titles and snippets: where to get listed or reviewed.
• answerText: the full AI answer, ready to drop into a client report.

From the second run on, the final summary record carries a sinceLastRun line such as "+2 answers now cite you, -1 no longer cites you". For the sources gained and lost, paste last week's dataset ID into previousRunDatasetId and the Actor adds a delta record.

To hand it over, open the dataset and use Export to get a CSV, then drop it into your own report template. A number makes the point, such as "your competitor is cited in 6 of 10 answers, you are in 1", and the source list is the plan to change it.

About the samples

AI answers vary between requests, so a single sample can miss a citation that normally appears. Any value above 1 gives each source a stability score. Start with 2 as above, and move to 3 for the clients you report on most.

Set up your first client report: one run, up to $1.80, and a repeatable AI visibility baseline for the client this week.