How to Measure AI Visibility | TAMGIO

1. List your brand, competitors, and the engines that matter

Before writing a single prompt, decide what you're actually tracking: your brand name (and common variants), 3-5 direct competitors, and which AI surfaces your buyers actually use. Ongoing tracking in TAMGIO runs on three engines — OpenAI web_search (gpt-4.1-mini), Perplexity Sonar (always grounded), and Gemini grounded search. Google AI Overviews (via SerpAPI) is added in Audit Mode, for four measured surfaces in total. Each surface answers differently — a brand that dominates Perplexity citations can be invisible in Google AI Overviews, so track them separately before averaging anything together.

2. Write 30-40 prompts across five intents

A single "best productivity app" prompt tells you almost nothing; AI answers change from one attempt to the next, and buyers ask in many different ways. Build a list of 30-40 candidate prompts spread across five intents: best-of ("best CRM for small teams"), comparison ("X vs Y"), how-to ("how to choose a CRM"), local ("CRM providers in Berlin"), and pricing ("how much does a CRM cost"). Write them the way a real buyer would type them, not the way you'd write ad copy. In TAMGIO you can draft this list yourself or generate candidates with AI and approve the ones that match how your buyers actually search.

3. Run each prompt as an N-sample scan, not a single check

This is the step most visibility checks skip. AI answers are non-deterministic — the same prompt sent twice can return a different set of brands, in a different order, with different citations. A single check tells you what the model said once, not what it tends to say. Running each prompt N times per engine, and repeating the scan on a cadence — daily or weekly — turns a noisy single sample into a proxy measure you can actually trust and trend over time.

4. Read Mention Rate and Prompt Coverage first

Mention Rate is the share of valid, grounded samples where your brand appears, averaged across your prompt list — it's the headline number for "how often do I show up." Prompt Coverage is stricter: the share of prompts where your brand is mentioned at least once across all samples. A brand can have a healthy Mention Rate while Prompt Coverage exposes that it's completely absent from an entire intent, like pricing or local — that gap is usually the most actionable finding in the whole scan.

5. Check Share of Voice against named competitors

Share of Voice compares your brand's mentions to your competitors' mentions within the same answers, at the sample level: your count divided by your count plus every named competitor's count. It answers a different question than Mention Rate — not "do I show up" but "when the topic comes up, whose name does the model actually reach for." Track it per engine; a brand can lead Share of Voice on Perplexity and trail badly on Google AI Overviews for the same prompt set.

6. Verify Citation Rate and don't confuse "no surface" with "not mentioned"

Citation Rate is the share of valid samples where the AI answer links directly to your own domain — the strongest form of visibility, since it's a clickable reference, not just a name-drop. Read it alongside the other three, because a brand can be mentioned often but rarely cited. One honesty check matters here: Google AI Overviews doesn't appear on every search. When it doesn't, that's "no surface" — the panel simply wasn't shown — and it is not the same as your brand being absent. Reporting the two the same way quietly inflates or deflates the real picture, so only grounded, successful answers should count toward any metric, while errors and missing surfaces stay visible instead of being hidden inside an average.