AEO reporting: what an agency's AI visibility report should include
A useful AEO report answers one question for the client: are we being named in the AI answers our buyers see, and is it getting better because of what we did? Every section should serve that question, per engine and per market, with a range instead of a single number.
This guide is for agencies building an AI visibility report, and for clients deciding whether the report they get is any good. Stellarcast sells agency plans, so read the last section knowing our bias; the checklist above it applies whatever tool you use.
Report the prompt set and markets, visibility and share of voice per engine with a P25 to P75 range, the competitors named instead of the client, the pages engines cite, the gaps you found, the changes you shipped, and the lift of each change with a confidence label. Put AI referral traffic last, as supporting evidence.
What an AEO report should include
| Section | What it shows | Why the client cares |
|---|---|---|
| Scope | The tracked prompts, the engines, the markets and languages, and the date range | Without a fixed scope, no number can be compared month to month |
| Visibility and share of voice | Per engine and per market: how often the client is named (mention rate), and its share of voice against named competitors, each with a P25 to P75 range | AI answers vary between runs, so one reading can move with nothing changed |
| Competitors | Which competitors are named on the client's prompts, and where they gained | Clients buy against competitors, not against a benchmark |
| Citations | Which URLs the engines cite: the client's pages, third-party pages about the client, and competitors' pages | Shows which pages earn the mention and where off-site work is needed |
| Gaps | Buyer questions where a competitor is named and the client is not, ranked | Turns the report into next month's work plan |
| Changes shipped | What was published or changed, when, and who approved it | Links the work to the result and keeps an audit trail |
| Lift | For each change, the movement on the prompts it targeted against prompts it did not touch, labeled MEASURED, LIKELY or TOO-EARLY | Stops the report claiming credit for an engine update |
| AI referral traffic | Sessions from AI assistants in Google Analytics | Supporting evidence only: many answers name a brand without sending a click |
Report branded and unbranded prompts separately. If every prompt contains the client's name, being named proves very little; the unbranded questions ("best accounting software for freelancers") are where share is won or lost.
See what a per-engine report looks like on your own brand: get a free AI visibility audit.
What to leave out
- One blended "AI visibility score". The engines differ, and an average hides the one where the client is losing.
- Screenshots of a single answer. One lucky answer is not a trend; a tracked prompt set is.
- Traffic as the only proof. Referral sessions are real but partial, because many AI answers send no click at all.
- Lift without a comparison. A before and after chart cannot separate your change from an engine update. Compare against prompts the change did not touch, or label the result too early.
How often to report
Collect the data far more often than you report it. Because AI answers vary from run to run, a prompt checked once a month gives you one noisy reading; checked daily, it gives you a range you can trust. Send the client a monthly report, keep a weekly digest for your own team, and alert only when a movement is bigger than the normal range between readings. Anything shipped in the last few weeks belongs under TOO-EARLY until the data says otherwise.
White-label and branded reports
Agencies usually need four things from the tool underneath the report: their own logo on it, the option to remove the vendor's mark, a PDF that still works after a share link expires, and an export of the raw history when a client leaves, because many contracts require handing that history over. Ask a vendor to show all four before you build a reporting process on it.
How Stellarcast handles agency reporting
- Per engine and per market. Google AI Overviews, ChatGPT, Gemini, Perplexity, Google AI Mode and Microsoft Copilot are checked daily on every paid plan, with visibility and share of voice each shown with a P25 to P75 range. The client-facing snapshot report exports as a PDF; the lift ledger comes in the CSV or JSON export below.
- Agency ($599 a month). 220 tracked prompts pooled across client brands, your logo on client reports, an export of the measurement history (the daily series with its P25 to P75 range, and the lift ledger, as CSV or JSON), API access, and three pitch workspaces, each a snapshot reading of a prospect that stays live for 21 days and does not draw on your pooled prompts.
- Agency Scale ($1,499 a month). 550 pooled prompts, reports without the Stellarcast mark, and five pitch workspaces on the same terms.
- Lift with a confidence label. Each shipped change is tracked against prompts it did not touch and labeled MEASURED, LIKELY or TOO-EARLY.
- A weekly digest and alerts. A weekly roll-up per brand by email, and alerts by email or Slack.
See the plans, or read how clients should vet an AEO agency to see the report from the other side of the table.
See what AI says about you today
The free audit asks Google AI Overviews, ChatGPT, Gemini and Perplexity the questions buyers in your market ask, and shows where a competitor is named instead of you. No card, no commitment.
Get your free visibility auditFrequently asked questions
What should an AEO report include?
The prompt set and markets, visibility and share of voice per engine with a P25 to P75 range, the competitors named instead of the client, the pages engines cite, the gaps found, the changes shipped, and the lift of each change with a confidence label. AI referral traffic belongs at the end as supporting evidence.
How often should an agency send an AI visibility report?
Monthly to the client is enough for most retainers, as long as the data underneath is collected daily or close to it. AI answers vary from run to run, so frequent readings give a range you can trust, while the report itself stays readable.
Can I white-label AI visibility reports?
With some tools. On Stellarcast, the Agency plan ($599 a month) puts your logo on client reports and the Agency Scale plan ($1,499 a month) removes the Stellarcast mark. Ask any vendor whether reports export as a PDF and whether you can export a client's history if they leave.
What is the most important metric in an AEO report?
There is no single one, which is why a blended score misleads. Start with visibility rate and share of voice per engine on unbranded prompts, each with its range, then show which changes moved them and how confident you are.