Citino teamReport AI visibility to leadership once a month, on one page. Show how often ChatGPT, Google AI Overviews and Perplexity name you against your competitors, what changed and why, what you did about it and whether it worked. Put a margin of error on every number, and say plainly when a change is noise.
The numbers come from AI search analytics. This article is about turning them into a report leadership will trust.
What leadership wants to know
Leadership does not need every metric. It needs answers to four questions:
- Are buyers hearing about us? When they ask AI assistants what to buy, are we named?
- Against whom? Who is named instead of us, and is that gap closing?
- What are we doing about it? Which fixes shipped, and did they work?
- Is it worth anything? What do AI answers send us in visits and pipeline?
Each section of the report answers one of them.
The one-page report
| Section | What to show | Why |
|---|---|---|
| Headline | One sentence: the biggest real change this month | Leadership reads this and maybe nothing else |
| Position against competitors | Share of voice per engine for you and your top three competitors, with margins | Answers “against whom?” |
| Visibility trend | Visibility per engine, this month against last, with margins | Shows whether you are named at all |
| What moved | The question groups that gained or lost, and the cited pages behind them | Turns numbers into causes |
| Fixes and verdicts | Each change shipped, the questions it targeted, its verdict | Answers “what are we doing?” |
| Visits from AI | Sessions from AI assistants and their conversions | Answers “is it worth anything?” |
| Next month | Two or three actions, each tied to a question group | Makes the report a plan |
Keep it to a page. Put the detail, such as every answer, every source and every question, in an appendix or a link to the tracking tool.
Show real changes, not noise
AI answers vary from run to run, so every number in the report is an estimate. Report each with a margin of error, calculated over questions rather than answers, and follow three rules:
- Inside the margin is no change. Label it within noise and do not call it a win or a loss.
- Check whether everyone moved. If a model update shifted every brand the same week, your change is not your doing. Compare with the median competitor.
- Never use one answer as evidence. A screenshot of a single ChatGPT answer is an anecdote. Leadership will remember it, and the next screenshot may say the opposite.
Worked example (illustrative). Your share of voice on Perplexity rose from 12% to 15%, with a margin of ±4 points. That is within noise: report it as “steady”, not “up 3 points”. On ChatGPT it rose from 9% to 17% with a margin of ±4, while the median competitor was flat. That is a real gain, and it belongs in the headline.
Report fixes with verdicts
Leadership funds work that shows results. For each fix shipped, report:
| Fix | Questions targeted | Verdict |
|---|---|---|
| The change, in one line | The question group, chosen before results came in | Confirmed, Negative, Inconclusive, Measuring or Awaiting first retrieval |
An inconclusive verdict is not a failure; it means the effect was too small to tell from noise with the data available. Reporting negative and inconclusive results alongside the wins is what makes the confirmed ones believable. How to tell whether an AI search fix worked explains each verdict.
Report visits honestly
AI assistants send some buyers to your site, and those visits show up in your analytics. In GA4, the Session source dimension shows the site that referred each session, so visits from AI assistants appear under their own sources. Google says AI Overview and AI Mode traffic is counted within overall search traffic in Search Console, so it cannot be separated there.
Report these visits and their conversions as a floor. A buyer who reads an answer, then searches your name or types your URL a week later, will not be counted as coming from AI.
Mistakes that cost credibility
| Mistake | Why it backfires |
|---|---|
| One blended “AI score” | Hides which engine and which competitor moved it |
| Reporting every wiggle | Next month’s noise reverses this month’s “win” |
| Changing the questions between reports | The trend breaks, and it looks like cherry-picking |
| Only reporting good news | Leadership stops trusting the good news |
| Claiming revenue from AI as a total | It is a floor; someone will find the buyers it missed |
A report that says “no real change” when nothing changed is the one leadership believes next time.
Where Citino helps
Citino is an AI search analytics tool. It asks ChatGPT, Google AI Overviews and Perplexity your buyers’ questions every day, with Gemini, Google AI Mode and Claude as add-ons and reports visibility, share of voice, position and sentiment per engine, each with the number of answers behind it and a 95% margin of error, as the methodology sets out. Changes inside the margin are shown grey, every fix gets a verdict, and AI visits come from your own GA4. Everyone tells you what to fix. Citino tells you whether it worked.