How to Measure AI Visibility for Your B2B Company

One of the practical challenges of AI visibility investment is measurement. SEO has a well-established measurement stack: Google Search Console for impressions and clicks, GA4 for session and conversion data, ranking trackers for keyword position trends. These tools work because Google search leaves a measurable trail when a buyer clicks through to your website from a search result.

AI-generated research does not leave the same trail. A buyer who uses ChatGPT to build a shortlist and then navigates directly to your website will appear in GA4 as a direct traffic visitor, with no attribution to the AI channel that drove them. A buyer who clicks through from a Perplexity citation may appear as referral traffic from perplexity.ai, but that is the only direct signal you get. The majority of AI-influenced research is invisible in standard analytics, which means you need a separate measurement approach if you want to track your investment.

The AI visibility measurement framework

Layer 1: Prompt testing and citation tracking

The most direct measurement of AI visibility is prompt testing: running a defined set of queries across ChatGPT, Perplexity, Gemini, and Claude on a regular cadence and recording which companies are cited in each response. This is manual but highly informative. A structured spreadsheet with your query set, the engines you test, and the responses you receive gives you a time-series dataset that shows whether your citation rate is improving, flat, or declining.

For each query, record: which companies appeared, whether your company appeared, what the response said about you (if anything), and which companies consistently appear without you. This tells you your citation rate (the percentage of relevant queries where your company is included in the AI response) and your competitive position within those responses.

Run this test at the same time each month, using a fresh session for each engine, with no prior context that might skew the results. After three to four months, you will have a trend line that clearly shows whether your AEO investment is moving the needle.

Layer 2: Referral traffic from AI engines

Some AI-influenced buyers do click through to your website from AI citations. Perplexity in particular cites sources and includes clickable links. Tracking referral traffic from perplexity.ai, chatgpt.com, gemini.google.com, and claude.ai in GA4 gives you a partial signal. Set up a custom channel group in GA4 that aggregates AI referral traffic as a single channel. This will undercount AI influence significantly (most AI research does not result in a direct clickthrough to your site), but it gives you a directional trend that can be tracked alongside your prompt testing data.

Layer 3: Share of voice in AI answers

Beyond tracking whether your company appears, track your share of voice relative to competitors. For each query in your prompt test set, count how many unique companies appear across all responses. Calculate your share: (queries where you appear) / (total queries) vs. the same metric for your main competitors. This gives you a competitive benchmark rather than an absolute number, which is more meaningful given the variability in AI engine responses.

A company with 15% AI share of voice in their category that was at 2% three months ago is showing clear progress, even if the absolute number seems small. A company tracking only whether they appear or not misses the competitive trajectory.

What to report to stakeholders

The metrics that translate best for marketing leadership and executives are:

• AI citation rate: Percentage of your tracked queries where your company appears in at least one engine response. Target: increasing month over month.
• AI share of voice: Your citation rate relative to the top competitors appearing in the same query set. Shows your competitive position.
• Engines covered: How many of the four major engines (ChatGPT, Perplexity, Gemini, Claude) cite your company for at least 50% of your tracked queries. A quality signal, not just quantity.
• AI referral traffic trend: Month-over-month change in sessions from AI engine domains. A revenue-adjacent signal that ties AI visibility to website activity.

A note on data variability

AI engine responses are not deterministic. The same query run twice in the same session can return slightly different results. This variability is real and is one of the reasons prompt testing requires a consistent, structured approach. Running each query three times per session and recording whether you appeared in at least two of three responses gives you a more stable citation signal than a single data point. The variability also means that single-month changes are less meaningful than three-month trends. Judge progress over a quarter, not a single test run.

For a 30-minute self-assessment you can run today, see how to audit your AI visibility in 30 minutes. For the ROI analysis that puts these metrics in a business context, see whether AEO is worth it for a B2B company.