Lead with the reporting question
Choose the decision the client needs to make: establish a baseline, investigate a source gap, review completed changes or approve the next experiment. A report assembled around that decision is more useful than an unexplained score.
Start from the editable report template if you need a lightweight structure before working in a client project.
Show what was planned and what was collected
Include the approved question set, market, surface, reporting period, valid answers and missing runs. State whether the counting unit is an answer, a brand mention or a source link. These are not interchangeable.
Measure AI visibility explains how to keep coverage separate from a mention rate and how to report an undefined denominator.
Attach findings to evidence
A useful finding names the question, observed answer and relevant source. Separate the client’s own citation from an independent citation and keep unverified interpretation out of the evidence column.
For each priority finding, attach the action owner, status and acceptance criteria. Avoid claims of guaranteed rankings, a fixed share of all user prompts or causation from a before-and-after sample.
Compare periods only when the method supports it
Record changed questions, aliases, competitors, market or collection method. If two periods use different populations, explain the break rather than presenting a seamless trend.
A client may still need a decision when coverage is incomplete. Show the missing portion and describe the narrower conclusion the available observations support.
Keep the delivery loop connected
The product links client context, questions, answer evidence, assigned work and report output. Verify the current export and collection availability in your workspace. Do not assume an unlisted engine, white-label capability or integration is available.
Explore agency delivery or inspect the illustrative demo before choosing a reporting workflow.