What should a monthly AI visibility report include?

How often each assistant recommended the client (X of Y answers), who took the spot when it didn't, the pages the assistants cited, and last month's numbers next to this month's.

Why not report a ranking in ChatGPT?

Answers change from run to run. AirOps found only 2.3% of citations stayed the same across 3 identical runs. Ask each question many times and report the share instead.

How do I show before and after?

Re-run the same questions on the same assistants the same number of times each month, then compare the share per assistant and check whether the pages you pitched now name the client.

Guides · Updated Oct 2, 2026 · By Rajan Chavada, founder of Soarce

How to report AI visibility to clients every month

What to put in a monthly AI visibility report for clients: recommended X of Y per assistant, who took the spot, the cited lists, and before and after re-runs.

The short answer

Report three things each month: how often each assistant recommended your client, who took the spot when it didn't, and which pages the assistants read to decide. Then show the same numbers from last month next to this month's. That's the whole report. Clients follow it because every number links to a real answer they can check.

Demand for this is real. AgencyAnalytics found 66% of agencies see rising client demand for AI search work (2026 benchmarks). Most clients don't yet know what a good report looks like, so a clear one sets you apart.

What to measure

Recommended X of Y, per assistant. For example, "ChatGPT recommended you in 6 of 80 answers." Show ChatGPT, Google AI Mode, Gemini and Perplexity apart. They read different sources: Perplexity and ChatGPT share only 11% of cited domains (Averi, a vendor).
Who took the spot. The brands recommended when your client wasn't, with a count for each. This often surfaces a rival the client never listed.
The cited lists. The roundups, reviews and comparison pages the assistants pointed to when they picked someone else. These are your outreach targets for next month.
Category vs brand questions. Keep "best X for Y" questions apart from questions that name the client. Brand questions look good and hide the gap.

Skip rankings. Answers change from run to run, so a single position means little. AirOps found only 2.3% of citations stayed the same across 3 identical runs (via SparkToro). Ask each question many times and report the share.

A sample monthly report outline

Section	What goes in it	Example line
1. Headline	One share for all questions, with last month next to it	Recommended in 14 of 250 answers (last month 6 of 250)
2. By assistant	X of Y for each assistant	Perplexity 8 of 50, ChatGPT 4 of 100, Gemini 2 of 50, AI Mode 0 of 50
3. Where you're missing	Question groups with 0 recommendations, and who won them	"Tools for small teams": 0 of 30, won by Brand A (9)
4. Who took the spot	Top rival brands with counts	Brand A 38, Brand B 21, Brand C 9
5. Pages to get on	Cited lists, ranked by how many rivals they name	A "best tools of 2026" roundup that names 6 rivals
6. What we did	Pitches sent, pages updated, listings won	Pitched 5 list editors, got onto 2
7. Next month	The next 5 pages to pitch	3 roundups and 2 review sites

Example numbers with made-up brands.

How re-runs show before and after

Run the same questions, on the same assistants, the same number of times, each month. If anything changes in the setup, the comparison breaks. Then put the two months side by side: the share per assistant, the rivals, and whether the pages you pitched now name your client.

Expect some lag. Perplexity can pick up new pages within 1 to 3 days on uncontested queries (Search Engine Land fake-brand test, Mar 2026). Other assistants can take longer. Freshness also matters on your side: AirOps found pages left 3 or more months without updates are 3 times likelier to lose AI visibility.

Tell clients up front that a small brand may start near zero. That's normal. When we studied 5 small SEO agencies, they were recommended in 14 of 1,119 "best agency" answers. See the agency study.

Doing this with Soarce

Soarce builds this report for you. Each scan asks ChatGPT, Google AI Mode, Gemini and Perplexity the client's buyer questions many times, and the report covers sections 1 to 5 above. See a sample report.

The agency plans are $49 a month for 10 branded client scans, or $149 a month for 25 scans plus monthly re-runs with before and after.

Questions

What should a monthly AI visibility report include?: How often each assistant recommended the client (X of Y answers), who took the spot when it didn't, the pages the assistants cited, and last month's numbers next to this month's.
Why not report a ranking in ChatGPT?: Answers change from run to run. AirOps found only 2.3% of citations stayed the same across 3 identical runs. Ask each question many times and report the share instead.
How do I show before and after?: Re-run the same questions on the same assistants the same number of times each month, then compare the share per assistant and check whether the pages you pitched now name the client.
Can Soarce build these reports for my agency?: Yes. Agency plans are $49 a month for 10 branded client scans, or $149 a month for 25 scans plus monthly re-runs with before and after.

Run a free scan