The Score That Isn't the Answer
Every AI visibility tool sells you the same headline metric: a discovery score, a visibility score, a mention rate. The pitch is that if the number goes up, business impact follows.
That's a leap of faith. And for a lot of brands, the faith isn't paying off.
We've seen this pattern across dozens of dashboards now. A company hires an AEO consultant, does the work — schema markup, FAQ pages, comparison content — and their AI visibility score climbs from 32% to 58% in six weeks. Legitimately great progress. On paper.
Then they look at their actual traffic and find that ChatGPT referrals grew maybe 3%. Perplexity referrals stayed flat.
Being recommended by AI isn't the finish line. Getting the click is.
Why Recommendations Don't Convert to Clicks
There are four ways a rising mention rate can fail to produce rising traffic — and they're all fixable once you know which one you're dealing with.
The mention doesn't include a link. When Perplexity or ChatGPT names your brand in a conversational answer without hyperlinking it, users have to break the conversation, open a new tab, and Google you. That's a real fraction of them; a lot don't bother. The mention counted. The click didn't happen. The link points at a broken canonical. AI models cache URL knowledge from months-old training data. If your URLs have changed — a rebranded product page, a legacy blog post that now 301s to a category — the redirect chain is a click killer. Some users bounce. Some just close the tab. The landing page doesn't match the query intent. A user asks ChatGPT for the best CRM for solo founders. Your CRM shows up in the answer. They click through to your homepage, which pitches Enterprise. The answer promised a fit; the landing page broke that promise. Bounce. Your brand is mentioned generically. "Some good options include X, Y, and Z" mentions everyone with equal weight. When AI leads with one product and adds you as a hedge, the click goes to the one it led with. You're in the answer, but you're not the answer.None of these problems show up in a visibility score. All of them show up in traffic — if you're looking at both.
The Correlation Nobody's Showing You
Here's the analysis that should be standard for AI visibility work and mostly isn't: plot your visibility over time against your AI-referred traffic over time.
Two lines. One dashboard. Same window.
When both lines move together, your AEO investment is compounding. When visibility rises and traffic doesn't follow, something in the click path is broken and you have a diagnostic exercise on your hands — not a mystery.
Most AEO tools stop at the visibility line. Most web analytics tools stop at the traffic line. The insight lives in the join, and nobody's joining them for you.
Four Divergence Patterns Worth Naming
Once you're looking at both lines, four patterns become useful vocabulary:
Visibility up, traffic up. The AEO work is landing. Keep doing what you're doing. Nothing to fix; just accelerate it. Visibility up, traffic flat. Warning shot. Something in the click path is bleeding. Audit the top-cited pages first: are canonical URLs consistent, is the query intent honored on the landing, is the mention accompanied by a link? Visibility up, traffic down. Emergency. You're being mentioned more but visitors are actively falling. This usually means a landing page changed for the worse (bounce triggers, load time, above-the-fold copy shift) or a redirect chain broke. Visibility down, traffic up. Rare but interesting. Existing mentions are converting at a higher rate than they used to. Look for what changed — a new hero section, a checkout redesign, a pricing update — and consider whether the visibility drop is temporary noise or a real signal to investigate later.These aren't sophisticated observations. They're the basic pattern-matching any performance marketer would apply to a paid channel. The problem is that AI visibility work has been happening in a silo where nobody's had the paired data to run the analysis at all.
The Cited-But-Not-Visited Optimization Queue
The most actionable output of joining visibility and traffic data isn't a chart — it's a list.
Your AEO efforts cause AI models to cite specific pages of yours in their responses. Some of those cited pages catch AI-referred traffic and some don't. The ones that don't are your highest-leverage optimization queue: the mention already exists, the click just isn't happening.
Concretely, this queue looks like:
| Cited page | Times cited (30d) | AI sessions (30d) |
|---|---|---|
| /pricing | 12 | 47 |
| /product/feature-x | 8 | 3 |
| /about | 4 | 0 |
| /blog/how-to-y | 3 | 0 |
You can't produce this list without both signals. AEO tools don't have your traffic. Traffic tools don't have your citation data. This is the join.
What Actually Measuring Both Looks Like
The mechanics are less complicated than they sound.
From the AEO side, you need: per-scan tracking of which URLs AI cites when it references your brand. Most serious AEO tools capture this; if yours doesn't, that's a red flag. From the traffic side, you need: sessions grouped by AI-source referrer (chat.openai.com, perplexity.ai, gemini.google.com, chatgpt.com). Any modern web analytics tool can produce this — GA4, Plausible, Mixpanel all support it. If you're on Shopify, Shopify Analytics does it natively via ShopifyQL. Then you need the join: a view that overlays the two on the same timeline, calls out divergences, and produces the cited-but-not-visited list. That's what Foxish's Traffic Impact page does — it connects to whatever analytics tool you already use and does the correlation math automatically.If you're not going to use Foxish for it, you can do the same analysis manually with a spreadsheet and thirty minutes of SQL against your GA4 export. The exact tool matters less than the discipline of looking at both signals together.
Stop Selling Yourself the Visibility Story
There's a soft trap in this space where AEO becomes its own KPI. Visibility scores go up, dashboards look green, and the harder question — "did any of this drive revenue?" — gets deferred to next quarter.
Some AEO work drives revenue. Some doesn't. The only way to know which is to measure both sides of the funnel and hold your work accountable to the click, not the mention.
The good news: if you've been putting real effort into AEO, the click is probably closer than you think. Fix the canonical URLs, tighten the query-intent match on your top-cited landing pages, and the traffic that's already been earned starts landing. The mention was the hard part. The click is the fixable part.
Try It on Your Own Data
Foxish measures both sides of the funnel out of the box — visibility across ChatGPT, Perplexity, Gemini, and Google AI Overviews, plus AI-referred traffic from GA4, Shopify, Mixpanel, or Plausible. The Traffic Impact page shows the correlation, calls out divergences, and produces the cited-but-not-visited queue for your brand automatically.
Free plan gets you started with the visibility scan. Growth ($99/mo) unlocks the Traffic Impact page. Pro ($179/mo) adds traffic-informed recommendations and divergence alerts.
Start a scan →