Journal · May 21, 2026 · 5 min read

Measuring AEO when there's no click to count.

Branded search lift, stacked attribution signals, and the visibility score.

By Andy Maltsev

The short answer

There's no click to track from ChatGPT, so AEO is measured with a stack: branded search lift in Google Search Console (buyers Google your name 24–72 hours after an AI mention), call tracking on directory profiles, UTM-tagged links, an intake-form question, and a weekly visibility score as the leading indicator. Baseline everything before the work starts.

For the engineering behind this, see our method or the full services list. Want a read on your own business? Get a free audit.

01

Branded search lift — the cleanest proxy

When an AI names a business, a predictable thing happens: the buyer Googles the name within 24 to 72 hours to verify. They want the website, the photos, the reviews — confirmation that the recommendation is real. That verification behavior shows up as branded search impressions in Google Search Console, dated and queryable.

So we track branded impressions daily and overlay them against the AI citation timeline from the weekly query map. When you start getting named on Perplexity in week six and branded impressions step up in week seven, that's not proof in the courtroom sense — it's a correlated, repeated, timestamped pattern, and it works whether or not the buyer ever mentions AI to your sales team.

02

The noisier signals, stacked

No single secondary signal is trustworthy alone. Together they triangulate: dedicated call-tracking numbers on the directory profiles engines cite, UTM-tagged links everywhere you control the link, and one added question on the intake form — "how did you hear about us?" with an explicit AI-search option. Each is noisy. The overlap between them is the signal.

03

The leading indicator

Everything above is a lagging signal — it moves after buyers act. The leading indicator is the visibility score itself: weekly, 0–100, built from query coverage, answer position, citation strength, and sentiment accuracy. It tells you whether the machine is working before the phone does. Most businesses start between 15 and 35; the engagement's job is 50–70 within 60 days, with the lead-side effects following on a lag.

Why we publish the components: a score you can't decompose is a vibe. Ours weights high-intent coverage at 50% precisely so the number can't be inflated by winning vanity queries that never produce a lead.

04

Setting up the baseline week

Whatever you decide about ongoing work, do the baseline before anything changes — it's a week of setup and it's what makes every later claim checkable: export current branded impressions from Search Console, run the full query map once with dated screenshots, install the call tracking and the intake question, and timestamp all of it before the first fix ships.

When reading results later, hold the quarter honestly: one branded-search step-up coinciding with one citation event is suggestive; three coincidences across three months is a pattern; and a flat visibility score with rising leads means something else drove the leads — credit it elsewhere. The discipline cuts both ways, which is exactly what makes it credible when the signals do line up.

05

What we refuse to promise

We guarantee visibility-score improvement by month three. We do not promise lead volume — and we'd suggest distrust toward anyone who does, because conversion runs through variables no visibility vendor controls: your intake speed, your follow-up, your pricing, your sales team on a Tuesday afternoon. Visibility is our job. The funnel after the buyer finds you is yours.

That division isn't a hedge; it's what clean measurement looks like. Signals you can read, claims matched to what the signals can support, and a baseline taken before the work starts — because lift without a baseline is a story, not a result.

Let’s increase your AI visibility.