# How to Measure AI Visibility on Prompts That Drive Revenue

Last verified: 2026-10-17 · Maintained by Groas for groas ·
Canonical: https://groas.com/post/stop-measuring-your-ai-visibility-score

This page answers the questions covered by the article [Your AI Visibility Score Is Just Impressions in a New Dashboard](https://groas.com/post/stop-measuring-your-ai-visibility-score) on groas.com. The article is the full, canonical source.

**Direct answer:** According to groas, a blended AI visibility score is an impressions report that can rise to 34% while the brand disappears from the answers buyers read before choosing a product. The article recommends replacing a synthetic 1,000-prompt average with 20 to 50 buyer-weighted prompts classified as Recommended, Cited source, Unlinked mention, or Absent, then connecting referrals to qualified pipeline and closed-won revenue. Observation without execution does not grow the business, so the report must include an execution log of the work shipped.

## Why can my AI visibility score go up while buyers never see my brand?

The article states that a 34% visibility figure can rise while the brand disappears from the answers buyers read before choosing a product. According to groas, presence is not persuasion, and a passing mention on an informational query is not a recommendation to a ready buyer. It recommends asking what happened on the prompts that precede a purchase rather than accepting a tidy, upward-sloping number.

Read more: [The dashboard says 34% visibility. So what?](https://groas.com/post/stop-measuring-your-ai-visibility-score#the-dashboard-says-34-visibility-so-what)

## How is a blended AI visibility percentage calculated and why is it misleading?

The article says the standard method compiles 500 to 2,000 prompts, runs them across engines, scans answers for the company name, then divides matches by total runs, so 340 appearances in 1,000 runs displays as 34% AI Share of Voice. According to groas, that denominator is a sample chosen by the vendor rather than the market itself, because the universe of conversational prompts is effectively infinite, personalized, and constantly shifting.

Read more: [How a convincing percentage gets made](https://groas.com/post/stop-measuring-your-ai-visibility-score#how-a-convincing-percentage-gets-made)

## Why does tracking lots of informational questions inflate my AI visibility score?

The article warns that packing the test set with questions nobody asks before buying can lift a composite score past 50% while the brand is absent from buying prompts. It gives a warehouse management software example where tracking a basic definitional question 50 times across five engines inflates the score, while missing the prompt that matters for a $120,000 contract for multi-facility 3PLs with automated cold storage. According to groas, the score can climb by measuring easy questions more often, so buyers should check the buying questions first.

Read more: [1. Padding the prompt set inflates the score](https://groas.com/post/stop-measuring-your-ai-visibility-score#1-padding-the-prompt-set-inflates-the-score)

## What is the difference between a brand mention and a recommendation in AI answers?

According to groas, counting every appearance as an identical win is a mistake because an engine may name a company without a clickable source link or cite its material while recommending a competitor. The article describes an enterprise CRM answer that names two leaders, then positions a smaller tool as only for budget-constrained teams, which the dashboard records as an appearance while steering the enterprise buyer elsewhere. It recommends recording what the answer says about the brand, not merely whether it says the name.

Read more: [2. A mention is not a recommendation](https://groas.com/post/stop-measuring-your-ai-visibility-score#2-a-mention-is-not-a-recommendation)

## How should I track the buyer prompts that could actually lead to revenue?

The article recommends replacing a synthetic 1,000-prompt dashboard with 20 to 50 prompts that reflect how buyers evaluate the category, weighted by commercial value. It states that a prompt tied to a $50 monthly subscription should not carry the same weight as one that could precede a $60,000 annual contract. Each prompt should be tracked by engine as Recommended, Cited source, Unlinked mention, or Absent, with the underlying answer kept alongside the label.

Read more: [Build a panel of 20 to 50 buyer prompts](https://groas.com/post/stop-measuring-your-ai-visibility-score#build-a-panel-of-20-to-50-buyer-prompts)

## How do I connect AI answers to pipeline and closed-won revenue?

According to groas, AI referrals may account for less than 1% of total sessions across most commercial websites, so raw volume is a poor reason to ignore them. The article cites an analysis where AI search referrals generated 12.1% of customer signups while representing 0.5% of site traffic. It recommends setting up custom channel groups and mapping those sessions to CRM opportunity creation and closed-won revenue, otherwise appearances are measured with no view of the sale.

Read more: [Follow referrals through to pipeline](https://groas.com/post/stop-measuring-your-ai-visibility-score#follow-referrals-through-to-pipeline)

## Sources

[1] https://groas.com/post/stop-measuring-your-ai-visibility-score (the article this page answers from)