How do I report AI search visibility to clients?
Quick Answer: To report AI search visibility, give each client a fixed prompt list, run it on a schedule, and report the same six metrics each period. Lead with share of voice for competitive SaaS clients. Lead with visibility when a client needs proof the brand appears. This affects any agency or consultant who owes a client a recurring AI visibility report.
Key takeaways
- Freeze each client's prompt list on day one of the period, and park new prompts in a second set.
- Give Claude and Perplexity their own weekly line in every chart you send.
- Recompute share of voice by hand before the report leaves your outbox.
- Close each report with the measure you will track next period.
How do you report AI search visibility to a client?
You report AI search visibility by asking AI engines the same buyer questions (prompts) every period and showing the client how the answers treat their brand.
Six metrics carry the report: visibility, share of voice, position, sentiment, sources and competitors. Welt Analytics reports all six for each tracked brand.
Welt Analytics checks ChatGPT, Gemini, Google AI Overviews and Google AI Mode daily, and Claude and Perplexity weekly [2], so the report carries two lines, not one blended trend.
In June 2026, Google's Search Central blog announced: "Today, we are launching a separate view dedicated to visibility from generative AI features." [1] Ask for Search Console access at kickoff.
What does AI search visibility reporting include?
AI search visibility reporting includes what the engines say about the brand and which other brands they name alongside it. Neil Patel's guide describes AI visibility tools as tools that "track brand mentions and content citations across LLM platforms like ChatGPT, Perplexity, Claude, and Google’s AI Overviews." [3]
When a metric does not move:
- Visibility at zero: the engines do not yet link the brand to that prompt. Keep watching it.
- Share of voice flat while visibility rises: competitors gain at the same rate.
- Position slipping while mentions hold: still named, but lower.
- Sentiment neutral: normal for a lesser-known brand. Quote the wording.
- Sources unchanged: list the top five pages the engines lean on.
- Competitors recurring: name them. See how to show a client which competitors AI assistants recommend over them.
Share of voice is your mentions / all brand mentions × 100. Semrush's guide writes it as "AI share of voice = (your AI mentions ÷ total AI mentions across all brands in your category) × 100" [4]. Before sending, recompute it from the report's counts; it should equal the printed percentage. If the brand drew 12 mentions out of 60, share of voice is 20%.
If it differs, your tool may weight position: Semrush's Brand Performance report "calculates AI SoV for you based on both how many times your brand is mentioned and how high it appears" [4]. On a count-only tool, a dropped prompt or engine is the likely cause.
Which metric should I lead with for each type of client?
Lead with the metric that answers the question the client asked when they hired you.
| Client situation | Top of the report |
|---|---|
| Brand-new client with no baseline | Lead with visibility on the first prompt set and label the first report as the baseline. |
| Competitive SaaS client | Lead with share of voice and the competitors that appear in answers, then position. |
| E-commerce client | Lead with visibility on product-category questions, then the sources the engines cite. |
| Client worried about how the brand is described | Lead with sentiment, quote the answer wording, and list the sources behind it. |
How do I build a client report you can repeat every month?
You build a repeatable report by fixing the inputs once and running the same steps every period. For tracking, see how agencies track client brands across ChatGPT, Gemini, Claude and Perplexity.
- Write the client's prompt list from buyer questions in their category, and freeze it for the period.
- Run the list once and save the results, with the date, as the baseline.
- Schedule the runs: the four daily engines every day, Claude and Perplexity once a week.
- At the end of the period, pull the six metrics for each engine group.
- Write the one-page summary: period dates, engines and cadence, the lead metric, the other five, competitors, top sources and next period's measure.
- Before sending, rerun three of the client's prompts by hand and compare the answer and the sources with the report.
Expected result: The brand mention, the cited sources and the competitors in the report match what the engine shows now. A mismatch can be normal run-to-run variation (see the third mistake below), a stale prompt or a changed answer; note it in the report instead of editing the numbers.
Google's AI Mode help page says: "For the best experience, enable history and personalized recommendations." [5] Rerun signed out, since a signed-in check can come back personalized.
To speed up step 1, Welt Analytics offers a 7-day free trial with no card required. It scans the client's site and sets the first prompts, so you start from a draft list.
What mistakes make an AI visibility report misleading?
Four mistakes mislead, and each has a pre-send check.
The first is changing the prompt list mid-period. Freeze the list and add new prompts as a separate set. To check, the prompt count at the start and end of the period should be equal.
The second is plotting daily and weekly engines on one line. Show Claude and Perplexity as their own weekly line and the four daily engines separately. Even the two Google features differ. Google's documentation says "AI Mode and AI Overviews may use different models and techniques, so the set of responses and links they show will vary." [6] To check, read every chart legend. Each line should name its engine and cadence.
The third is reading one run as the whole picture. Semrush writes that the platforms return different responses to the same prompt within a single day, and that a share of voice swinging between 20% and 40% over a day is normal [7]. Report the period's pattern across scheduled runs, not a single answer. To check, each number in the report should say how many runs it covers.
The fourth is promising the client a rank or a traffic result. Semrush notes that traffic can stay flat while brand visibility grows significantly [8], so a traffic promise can fail even when the work succeeds. Report what was measured and promise no outcome. To check, search your draft for "will" next to a rank or traffic figure. You should find none.
Which tool setup makes client reporting easier?
A tool makes client reporting easier when it holds a prompt set per brand, covers the engines you report on, records sources and competitors, and exports data. For onboarding speed, see which GEO tool is easiest to set up for a new client in one day.
Welt Analytics plans; listed prices exclude VAT, price as of November 13, 2026 [2]:
- Starter: 89 EUR a month, 50 prompts, 1 brand, all six engines, unlimited seats.
- Growth: 229 EUR a month, 150 prompts, 2 brands, plus a Data API.
- Pro: 499 EUR a month, 400 prompts, 3 brands, plus a Data API.
- Enterprise: custom prompts and brands, with pricing agreed individually.
More client brands means a higher plan. If you already run Semrush, plan around its tier structure: Neil Patel's review notes that "Semrush limits how many projects you can track per account tier." [3]
To run one fixed prompt list per client across ChatGPT, Gemini, AI Overviews and AI Mode daily and Claude and Perplexity weekly, start with Welt Analytics.
Questions
How often should I send an AI visibility report to a client?
Monthly is a sound default. Daily engines give you about thirty readings a month. Claude and Perplexity give you four or five, so flag that their line rests on fewer points. Once you pick a period, keep it.
Can I export AI visibility data for my own client dashboard?
Some trackers allow it. Slate's page says users can "Download raw result data to CSV for custom reporting or deep-dive analysis." [9] On Welt Analytics, the Growth, Pro and Enterprise plans list a Data API. Check the plan before you promise a client a custom dashboard.
What do I tell a client when their share of voice is low?
Call it the starting baseline. Name the competitors that take the mentions, show the sources the engines used, and agree on the figure you will compare against next period. Make no forecast of a rise.
Do I need a separate prompt set for every client?
Yes. A shared list blends two brands' mentions into one count, so neither client's share of voice is accurate. Count your client brands against your plan's brand limit before signing the next one.
Sources
- Google Search Central Blog, "Gen AI performance reports", June 2026
- Welt Analytics, home page
- Neil Patel, "AI visibility tools"
- Semrush, "How to measure AI share of voice"
- Google Search Help, "AI Mode"
- Google Search Central, "AI features and your website"
- Semrush, "How we are using Semrush to drive LLM visibility"
- Semrush, "How to measure AI visibility"
- Slate, "AI visibility tracker"