How to Report AI Visibility to Clients (Agency Guide)

Last updated: October 2026

A good monthly AI visibility report shows five things: the score trend, the mention rate per engine, the gaps where a competitor is named and the client is not, the citations you won, and share of voice against named competitors. Add one page on what you shipped and what comes next, and explain variance before the client asks.

Key takeaways🔗

• Report trends over at least three months, not single readings. AI answers change from run to run, so one week tells you very little.

• Lead with gaps and wins. Clients care more about "a competitor is recommended for this question and you are not" than about a score.

• Do not report a "rank position" in AI answers. Independent research found list order changes almost every time a prompt is run.

• Pick a white-label format your client will actually open: a branded PDF, a share link, a Looker Studio report or a client portal.

• Tie every gap to a fix you shipped or will ship. A report with no action list reads like a bill.

What should a monthly AI visibility report include?🔗

A monthly AI visibility report should answer one question for the client: are AI engines recommending us more or less than last month, and what are you doing about it? Five metrics cover that, plus an action page.

If you need a refresher on how the headline number is built, our explainer on how an AI visibility score is calculated walks through it, and our guide to AI share of voice covers the competitor comparison.

How do you show the score trend without overselling it?🔗

The score trend is the first chart a client sees, so show it over time and resist the urge to celebrate one good week. Plot weekly readings as dots and draw a line through the monthly average. That way a client sees both the movement and the noise.

Write one plain sentence under the chart. For example: "Your visibility rose from the low 20s to the high 20s over three months, mostly from new Perplexity mentions on comparison prompts." Avoid exact decimal changes; they suggest more precision than the data has.

For the mechanics of building a reliable trend line, see how to track ChatGPT performance over time.

What is mention rate, and why report it per engine?🔗

Mention rate is the share of tracked prompts where an engine names the client at all. It is the simplest number a client can understand: "ChatGPT named you in 9 of 25 buyer questions this month."

Report it per engine because the engines disagree. A client can do well in Perplexity, which pulls live web results, and poorly in ChatGPT on the same prompts. Averaging the two hides the problem you are being paid to fix. If you use TrendlyAI, the tracked engines are ChatGPT, Perplexity, Claude and Google AI Overviews. It does not track Gemini, Grok, Copilot or Google AI Mode, so say that in the report if the client asks about them.

How should you present gaps to a client?🔗

A gap is a buyer prompt where a competitor is recommended and the client is not, and it is the most useful line in the whole report. It turns an abstract score into a specific, fixable problem.

Do not list every gap. Pick 5 to 10, ranked by how close the prompt is to a purchase. "Best payroll software for restaurants" matters more than "what is payroll." For each one, show:

1. The exact prompt.

2. Which engines named which competitors.

3. The page or source the engine cited, if it cited one.

4. The fix you plan, such as a comparison page, an FAQ page, a Reddit answer or a product page rewrite.

Our guide to competitor AI visibility analysis shows how to find the source behind a competitor's mention.

What counts as a won citation?🔗

A won citation is a prompt where the client was missing on an earlier scan and is named or cited on a later one. It is the closest thing GEO has to a "ranking win," and clients love it.

Be careful with it. Because answers vary, a client can appear once by chance. Count a citation as won only when it holds on at least two scans in a row, and link it to the page or fix that likely caused it. TrendlyAI flags won citations on later scans, but the judgment about why it happened is yours to explain.

How do you explain variance to clients?🔗

Explain variance in the first report, before a bad week makes it look like an excuse. The short version: AI engines do not return a fixed list. Ask the same question twice and you can get different brands, in a different order.

There is good independent research to point to. In January 2026, SparkToro and Gumshoe had 600 volunteers run 12 prompts through ChatGPT, Claude and Google's AI Overviews 2,961 times. They found less than a 1 in 100 chance that ChatGPT or Google's AI would give the same list of brands in any two of 100 responses. The same write-up concluded that visibility percentage across many prompts and runs is a reasonable metric, but that a "ranking position in AI" is not (SparkToro, January 2026).

Three habits keep the conversation calm:

• Report rates, not positions. "Named in 40% of prompts" is meaningful. "Ranked third in ChatGPT" is not.

• Use more readings. More prompts and more scans narrow the noise. A monthly view built from four weekly scans is steadier than any single week.

• Be honest about your sample size. TrendlyAI runs weekly scans with 1 sample per prompt per engine. On 25 prompts and 4 engines, that is 100 readings a week and roughly 400 a month. That is fine for a monthly trend, but a 2 or 3 point weekly move is not a story.

When I set up TrendlyAI's own tracking, our score was low, and I publish it anyway. Showing a client a low starting number with a clear plan builds more trust than a flattering snapshot.

Which white-label options do agencies have?🔗

White-label options range from a branded PDF to a full client portal on your own domain. Pick the one that matches how your client reads reports, not the most features.

Prices are monthly in USD from each vendor's pricing page: Rank Prompt, Otterly.AI, SE Visible, Peec AI. We build TrendlyAI, so weigh that. Its weak points: only 4 engines, weekly scans with 1 sample per prompt, no prompt-volume data, few public reviews and a small team. Our full roundup of white-label AI visibility tools covers more options.

A rule of thumb: small clients rarely log into anything. A branded PDF in their inbox plus a share link they can forward is usually enough. Larger clients with in-house marketing teams often want a live dashboard or portal.

How long should the report be, and how should you deliver it?🔗

Keep the report to what a busy owner will read in five minutes: one summary page, one page of metrics, one page of gaps and actions. Anything longer belongs in an appendix.

A simple order that works:

1. Summary: three sentences. What changed, why, and what you will do next.

2. Metrics: score trend, mention rate per engine, share of voice.

3. Gaps and wins: top gaps, won citations with the page that earned them.

4. Actions: fixes shipped, fixes planned, anything you need from the client (approvals, product details, reviews).

Send it on the same day each month. If you sell AI visibility as a package, our post on GEO services for SEO consultants covers how to price and scope it, and the TrendlyAI agency page shows how client brands, white-label PDFs and pitch audits fit together.

How do you start reporting for a new client?🔗

Start with a baseline before you change anything, so every later report has something to compare against. Run the client's top buyer prompts, record which competitors appear, and save that first report as month zero.

The quickest way to get a first look is TrendlyAI's free AI visibility check: no signup, about 30 seconds, 5 prompts on ChatGPT and Perplexity. It is also a useful screen to share in a sales call. For every other guide on this topic, see our GEO hub.

Frequently asked questions🔗

How often should agencies report AI visibility to clients?🔗

Monthly works for most clients. Weekly data is useful for your own team, but a monthly report built from several scans is steadier and easier for a client to read. Report weekly only during an active test where the client is waiting on a result.

Should I report a client's ranking position in ChatGPT?🔗

No. SparkToro's January 2026 research found that AI tools rarely return the same list of brands twice, and the order changes even more often. Report how often the client is named across many prompts (mention rate) instead of a position.

What if the client's score drops one month?🔗

Check whether the drop holds across engines and prompts before you explain it. A small one-month dip is often normal variance. If it is real, look at which prompts were lost, which competitor replaced the client, and what that competitor's cited source is.

Can I white-label TrendlyAI reports?🔗

Yes. White-label PDF reports and share links are included on every TrendlyAI plan, from Solo at $19/mo to Agency at $149/mo for 10 client brands. Paid plans start with a 7-day free trial, no credit card.

What is the difference between mention rate and share of voice?🔗

Mention rate is how often an engine names the client across tracked prompts. Share of voice compares the client's mentions with those of named competitors on the same prompts. A client can have a rising mention rate and a flat share of voice if competitors are growing too.

Related reading🔗

• Best white-label AI visibility tools

• GEO services for SEO consultants

• What is AI share of voice?