How Do You Know If AI Is Recommending You? A Way to Actually Measure It

Say you’ve done the work — machine-readable pages, structured data, a clean presence on the sites AI trusts. Fair question: how would you even know if an assistant is recommending you? You can’t improve what you can’t see, and this is the part most people skip.

Start with the uncomfortable truth: most of your AI visibility is invisible in your analytics. Google Analytics won’t tell you that ChatGPT named your winery to a shopper who never clicked. Google’s own AI Overviews blend into your normal search traffic with no separate label. And even when an AI does cite you, most of those citations never turn into a click — research puts Perplexity’s click-through at roughly 12–18%. Your real AI influence is likely a few times larger than anything a dashboard reports. So if you’re waiting for a clean “AI traffic” number, you’ll wait forever.

Here’s how to measure it anyway, cheapest first.

1. Ask the machines yourself. (This is the ground truth.)

Open ChatGPT, Gemini, Perplexity, and Claude, and type the questions your buyers actually ask — “recommend a small-production Napa Cabernet,” “which wineries near Paso Robles sell allocations to the public,” “a good sparkling wine under $40.” Then note two things: were you named, and who was? This is the single most direct signal of whether AI recommends you, it’s free, and it’s exactly what your customers are doing. Do it once a month and keep a simple log.

2. Check your analytics for AI referrals.

When an AI answer does send someone to your site, it shows up in Google Analytics as referral traffic, not search. Filter your traffic sources for chatgpt.com, perplexity.ai, gemini.google.com, and copilot.microsoft.com, and you’ll see the clicks the assistants are actually driving. Small numbers are normal — remember, most citations don’t produce a click.

3. Use Search Console’s AI view.

Google added a way to see when your pages are cited in its AI answers inside the Search Console performance report. It’s incomplete, but it’s the closest thing to a direct readout from Google itself.

The metric that actually matters

Notice what changed: the goal isn’t “how much traffic did AI send me.” It’s “am I named when a buyer asks?” A recommendation you never get a click from still put you on someone’s shortlist — that’s the whole game. So track presence, not just clicks.

Your whole measurement system can be a spreadsheet: five buyer questions, asked monthly, in each assistant. Columns for “named? (yes/no)” and “who else showed up.” Watch the trend over a few months, not any single day. If your name starts appearing where it didn’t before, the work is landing — long before it ever shows as traffic.

You don’t need a paid tool to begin (they exist, and they help once you scale). You need to ask the questions your buyers ask, write down whether the machine says your name, and do it again next month. That’s more than almost any of your competitors are doing — which is exactly why it’s worth doing.

Zillah Bahar is the founder of COLAClear and writes about getting found by AI at GetDiscoverable.io.

Sources: Tracking AI search referrals & click-through rates — Foundry · How to track AI traffic in GA4 — SUSO · Monitoring Google AI Overview citations — Otterly