TL;DR Summary:
Modeled ROI: Semrush says AI visibility ROI should be calculated from modeled revenue minus costs, divided by costs, then multiplied by 100, because AI influence often shows up without a direct click.Track the Full Path: The framework combines prompt tracking, early signals, form fields, GA4 and CRM links, and sales rep feedback to catch revenue that analytics alone misses.Know the Blind Spots: GA4 undercounts AI traffic because some AI sessions are mislabeled or stripped of referrers, so direct referral data alone will underreport true AI impact.Separate Revenue Types: Semrush recommends splitting AI-driven revenue into direct, assisted, self-reported, and modeled buckets so teams can tell confirmed results from estimates.How do you know if showing up in ChatGPT answers actually makes you money? Semrush published new guidance this week laying out a formula-based approach to answering that question, and it matters because most companies still measure AI visibility like a traffic source, when it works more like an influence channel that rarely produces a trackable click.
Why AI Visibility ROI Needs a Different Formula Than SEO
The new guidance breaks AI visibility ROI into a specific calculation: subtract your AI visibility costs from the profit those efforts generated, divide by the costs, then multiply by 100. Semrush walks through a B2B example where 1,000 AI-attributed visits at a 4% lead rate produce 40 leads, 12 qualified leads at a 30% qualification rate, and roughly 2.4 closed deals at a $6,000 average contract value. That works out to about $14,400 in modeled revenue. Against $4,000 in monthly costs, the resulting AI visibility ROI comes out to 188%. The guidance is clear that this number is modeled, not observed, and should be labeled that way in any report.
The Five-Part Framework Behind Measuring AI Visibility ROI
Semrush recommends five steps: baseline prompt tracking, watching for early signals, adding AI options to intake forms, connecting GA4 and CRM data, and training sales reps to ask prospects how they found you. Each step catches a different piece of the picture. Forms and sales calls catch what analytics tools miss, since a buyer who reads an AI recommendation and converts weeks later through a different channel never shows up as an AI referral. Before any of that revenue math can be trusted, though, a company needs a real accounting of where it actually appears in AI answers relative to competitors and why, which is the visibility layer that tools like AI Mentions are designed to surface, since ROI formulas are only as good as the presence data feeding them.
Where GA4 Falls Short in Tracking AI Visibility ROI
Google Analytics 4 added an “AI Assistant” default channel group that automatically groups sessions from recognized AI referrers. The catch: it only classifies sessions from its rollout date forward, so older AI visits stay filed under their original channel. It also depends on the referrer data being passed along, so a click inside a native AI app that strips that information lands in Direct traffic instead. Perplexity currently gets grouped under Referral rather than AI Assistant, and traffic from Google AI Overviews still counts as Organic Search. Anyone calculating AI visibility ROI from GA4 alone will undercount the real number.
Separating Confirmed Revenue From Estimated Revenue
The guidance introduces four CRM buckets for classifying AI-driven leads: direct, meaning a tracked AI session that converted; assisted, meaning AI touched a longer multi-touch path; self-reported, meaning the buyer told you directly; and modeled, meaning the number comes from estimated signals rather than hard data. Keeping these separate lets a finance team see exactly which revenue figures are confirmed and which are estimates. Semrush’s ecommerce example shows this in practice: 1,000 AI-attributed visits at a 2% conversion rate and an $80 average order value produce $1,600 in revenue and, at a 60% gross margin, $960 in gross profit. Against $600 in monthly costs, that comes to a 60% ROI.
The one thing to watch going forward is whether your own reporting still treats AI visibility ROI like a click-based channel. If your dashboard only counts direct AI referral traffic, you are almost certainly underreporting what AI visibility is doing for your revenue. Build the habit of separating confirmed numbers from modeled ones, and keep the estimates conservative until your data catches up. Before you can model any AI visibility ROI, you need clarity on where you actually appear in AI answers versus competitors, which is exactly the gap AI Mentions is built to diagnose.
