An AI visibility report your client can use.
Show whether AI answers recommend the client, what sources they cite and what your team will do next. Start with the blank template, then use the example below to check your numbers.
Start with the client's question
“Does ChatGPT recommend us?” sounds simple. Your report needs to explain which questions you tested, where you tested them and what the answers actually said. A link to the client's website is useful evidence, but it doesn't necessarily mean the answer recommended the business.
Lead with a short answer the client can repeat to a colleague. Then show the counts behind it and the work you recommend. Keep the detailed answer records in the client's private workspace.
A usable report has six parts: an executive summary; test coverage; recommendations and competitors; citations and disclosed searches; traffic and enquiries where measured; and actions with owners and review dates. If you measure sponsored placements, put those in a separate paid section.
The template: six sections to fill in
The free Markdown template is a plain-text document you can open in a text editor or paste into your usual reporting tool. It has blank tables and calculation notes. It contains no customer data, and downloading it doesn't connect tracking or generate a report inside Reputably.
1. Executive summary
State the reporting period, market and the clearest finding. Explain whether the result concerns a recommendation, a brand mention or a citation. Give the client one decision to make and name any coverage gap that materially affects it.
2. What you tested
List the exact questions and variants, engines, market, language, dates and collection method. Show planned checks, usable answers, unavailable answers and errors separately. Keep Google AI Overviews separate from Google AI Mode. If the question panel or collection method changed, split the comparison at that point.
3. Recommendations and competitors
Read the answer before relying on a mention flag. Count explicit product or business recommendations separately from incidental name mentions. If you report position, explain how you order an answer without a numbered list; exclude answers where no defensible position exists. Record which competitors appear and the context in which they're recommended.
4. Citations and disclosed searches
Count each answer citing the client's site once, even if it contains several links. You can also list distinct cited pages, but label that as a separate measure. Note which passage was used and whether it describes the client, a competitor or general advice. Keep disclosed fan-out searches separate from your tracked questions. When no searches are disclosed, record “unknown,” not zero searches performed.
5. Traffic and enquiries
Include detectable AI referral visits only when analytics is configured and checked. Report crawler requests separately: a machine fetching a page is not a customer visit. Show enquiries or trials from the attribution method you actually use and explain what it misses. If tracking isn't installed, write “not measured.”
6. Actions and review dates
For each finding, name a concrete page or workflow to improve, its owner, the completion date and when you'll check it again. Link to the finished work. “Improve AI visibility” is too broad; “add service-area details to the emergency plumbing page, then recheck the same questions” gives everyone a task they can complete.
Worked example: a fictional local-service client
All names, questions and numbers in this example are invented. They demonstrate the calculations; they are not Reputably results or an industry benchmark. Imagine Harbour Plumbing, a fictional Sydney business, testing ten fixed buyer questions once per engine in each of two weekly windows.
| Engine | Usable answers | Unavailable | Errors | Recommendations | Answers citing the client's site |
|---|---|---|---|---|---|
 ChatGPT | 10 | 0 | 0 | 2 | 3 |
 Claude | 10 | 0 | 0 | 1 | 2 |
 Gemini | 10 | 0 | 0 | 3 | 4 |
 Google AI Mode | 10 | 0 | 0 | 2 | 3 |
| Google AI Overviews | 8 | 1 | 1 | 0 | 1 |
| Total | 48 | 1 | 1 | 8 | 13 |
The recommendation rate is 8 ÷ 48 = 16.7% of usable answers. The site citation rate is 13 ÷ 48 = 27.1%. Coverage is 48 ÷ 50 = 96% of planned checks. Recommendations and citations can overlap, so don't add them to produce a “total visibility” number.
The previous week had 50 usable answers and six recommendations. Simply comparing 12% with 16.7% would hide a coverage change. For a fairer comparison, match the same 48 question-and-engine checks available in both weeks. Suppose five of those matched prior answers recommended the client: the comparison becomes 5/48 (10.4%) versus 8/48 (16.7%), an increase of 6.3 percentage points. Show the unmatched checks separately and keep testing; this small example doesn't establish that an edit caused the change.
How to describe one answer
Suppose an answer links to Harbour Plumbing's blocked-drain guide but recommends two other plumbers. Record a site citation and no recommendation for Harbour Plumbing. A second link to the same guide in that answer doesn't create a second citing answer.
What the client summary could say
In this week's 48 usable answers, eight recommended Harbour Plumbing and thirteen cited our site. One check returned no overview and another failed. The blocked-drain guide is being used as a source, but some answers still recommend other plumbers. We'll clarify the areas we serve and add the information a homeowner needs to decide whether to call, then review the same questions next week. AI referral visits are not measured yet.
This summary gives the client a result, its limits and a next action without turning a handful of answers into a claim about every AI user.
Use Reputably to investigate the result
- Choose the client and reporting period. In the client's AI visibility view, check the tracked questions and market before comparing dates. Confirm the runs completed.
- Read an answer behind the score. Check whether the client is recommended, mentioned in passing or absent. Keep the original question and engine with your notes.
- Inspect the source links. Open the cited public page and find the passage the answer relies on. Does it explain the client's offer, or is the engine using it to discuss someone else?
- Use disclosed searches to choose useful work. A recurring question about pricing may justify clearer pricing. It doesn't justify publishing a near-duplicate article for every query variation.
- Fill in the report and assign the work. Use the template for your client summary, then revisit the same question panel after the change is live.
Reputably's AI visibility product page explains the answer surfaces and evidence available. The fair testing guide covers repeatable checks, while the agency workspace guide explains client organisation. Access and available evidence depend on the account and run; no new paid checks are needed just to use this document template.
Questions clients and agencies ask
What should an AI visibility report include?
Include an executive summary, test coverage, recommendations and competitors, citations and disclosed searches, traffic and enquiries where measured, and actions with owners and review dates.
Does a citation count as a recommendation?
No. An answer can cite a business's guide while recommending a competitor. Read the answer and report source citations separately from recommendations.
How should missing answers be counted?
Keep unavailable answers and errors outside the denominator of usable answers, and show them in the coverage table. Compare matching questions and engines across periods so missing checks don't disguise a change in the sample.
How often should we report to clients?
A monthly summary can work alongside more frequent checks. Agree on a cadence your budget supports, keep the question panel stable and review repeated observations before describing a trend.
Can we charge for this work?
You can scope research, reporting and content improvements as agency services. Agree on the work and measurement with the client. Don't promise a specific ranking or guaranteed recommendations in exchange for the fee.
Use this guide with the AI visibility tool comparison and Reputably for agencies. Template and worked example updated 25 September 2026.
Explore the ideas in practice
See how Reputably brings social listening, AI visibility and reviews together.