How to Measure AEO: The Analytics That Prove AI Is Sending You Customers

Everyone is selling Answer Engine Optimization right now. Almost nobody can prove it worked.

That is the dirty secret of the AEO gold rush. An agency tells you they will "get you recommended by AI," you pay for six months, and at the end you have a stack of deliverables and no honest answer to the only question that matters: is AI actually sending you customers?

We think you deserve a real answer. So here is the honest version — what you can measure, what you can't, and what to ignore.

Why AEO feels unmeasurable (and why that's a convenient excuse)

When someone finds you through Google, you get a clean trail. When ChatGPT recommends you in a conversation, the customer often just… shows up. They type your name into their browser. They call. The AI that influenced them never passes a tidy referral.

That gap is real. It is also the thing lazy agencies hide behind — "AI search can't be tracked, just trust the process." Don't accept that. The trail is fainter than Google, but it is absolutely there if you know where to look.

The five things you can actually measure

1. AI-engine referral traffic. AI tools do leak referrals — more every month. In Google Analytics 4 you can isolate visits coming from chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com and the rest. Segment them out and you have a hard, growing number: humans who clicked through from an AI answer to your site.

2. Search Console — including AI Overviews. Google Search Console shows the impressions and clicks behind the AI Overviews now sitting on top of search results. Which questions are you appearing for? Which are you winning? This is ground truth, straight from Google, and most businesses never look at it.

3. Share of AI voice. This is the one almost nobody runs, and it is the most direct of all: take the real questions your buyers ask — "best [your service] in [your city]," "who should I hire for X" — and put them to ChatGPT, Perplexity, and Gemini on a schedule. Are you named? Are your competitors? Track it over time and you have a scoreboard for exactly the thing AEO is supposed to move.

4. Branded-search lift. Here is the metric that catches the "invisible" conversions. When an AI recommends you, the next thing people do is Google your name. So when AI is working upstream, your branded search — searches for your company specifically — climbs, even when the referral itself never showed up. A rising branded-search line is one of the strongest signals that AI is putting you in front of people.

5. The leading indicators. Before any of the above moves, the inputs have to be right: structured data (schema) coverage, whether AI crawlers can actually read your pages, and page speed. AI engines favor fast, well-structured, machine-readable sites. These are the early warning lights — fix them and the downstream numbers follow.

The metric to ignore

Be skeptical of anyone selling you a single "AEO score" that always goes up and to the right. Some attribution in AI search is genuinely fuzzy — and a dashboard that pretends otherwise is selling you false precision, not truth. What you want is a trend line across the five signals above, read honestly, not one magic number engineered to make the invoice feel justified.

Measuring is step one. Acting on it is the whole point.

Numbers you never act on are just decoration. The reason we wire all five of these into one place — GA4, Search Console, share-of-voice checks, branded search, and the technical health of the site — is so the loop actually closes: see what's working, fix what isn't, measure again.

That is what the Nexus Studio does, and it is what ARIA does on top of it: she watches the analytics, spots the question where a competitor is getting named instead of you, and goes and does something about it — the content, the schema, the page. Measure, act, re-measure. That is the difference between an AEO report and an AEO result.

If you are about to spend real money getting recommended by AI, spend it where you can see the return.

See how the Nexus Studio tracks it, or learn how REINS keeps the engine running.