How agencies report AI search visibility to clients

A client report on AI search holds up when it has four parts: a fixed scope, a dated baseline, the same measurements each month, and a short list of next steps. The structure matters more than the tool that produces the numbers.

1. Scope

Write down the client, the market, the engines and the prompt set. Freeze the prompt set. Changing a question between months breaks the comparison, so if a question must change, record it as a new series.

2. Baseline

Run the frozen set once, before any work starts, and date the result. Every later number is read against this run.

3. Monthly change

Report the same few measures each month:

Measure	Baseline	This month	Note
Questions where the client is named	—	—	Which questions moved and why
Citations to the client’s pages	—	—	Which pages were cited
Answers that misstate the client	—	—	The wrong fact and its source

Fill the blanks with the client’s own numbers. Keep the notes short, and quote an example answer wherever a number moved.

4. Next steps

End with three actions, each tied to a finding in the table. A client can act on three actions. A list of twenty is a list nobody reads.