Measure AI visibility through phone calls: what 30 million CallRail data points reveal
Summarize this article with AI
A client calls me on a Thursday morning. « Stéphane, my traffic is down, but my sales are holding. »
I pull up his Search Console.
Yes, organic traffic dropped 11% in 6 months.
But average order value climbed 18%.
And inbound calls tripled.
His sales team is drowning in leads.
Yet the site gets fewer visits. Paradox.
We dig for 2 hours on a video call.
We activate data from a call tracker installed 3 months earlier.
And suddenly, the picture clears: 64% of new calls come from results generated by artificial intelligence in Google.
Not a bug.
A shift.
In just three years, the proportion of calls triggered by AI-generated results has nearly quadrupled. Here’s the trajectory CallRail measured across 30 million calls.
The rise of AI-driven calls: from 11% to 42% in 3 years
Share of inbound calls attributed to AI Search interactions (CallRail data)
30 million calls analyzed: what CallRail uncovered
CallRail analyzed 30 million inbound calls in e-commerce.
42% of calls trace back to an interaction with AI Search: snippet, local pack, AI Overview, conversational response.
In 2024, this figure was 11%.
In 2025, 28%.
Today, 42%.
Calls from AI arrive on average 22% faster after the search.
The customer didn’t click a link; they hit « Call » from the card or enriched block.
The classic click-through journey is being replaced by a direct call from the AI Overview. This flow shows the new, faster route Google’s conversational responses enable.
How the search-to-call path has shortened
From a multi-step funnel to a two-step tap
Why AI picks up the phone for you
A buyer types « best silent professional juice extractor ».
They see an AI Overview with 3 models, a built-in comparison, and a call button for each seller.
They tap the screen, not a click.
I observe that the classic path « query > click > product page > contact form > call » is dead.
Now it’s « query > AI response > call ».
Two steps skipped.
22% faster.
The content that feeds that AI response is your semantic cluster.
The one structured so entities are clear, comparisons trustworthy, trust signals obvious.
« AI doesn’t generate speech. It synthesizes what’s already there. And if it’s you who’s there, it’s your phone that rings. »
67 calls in one month: what a parts retailer implemented
A client sells parts for professional espresso machines.
Catalog of 1,200 SKUs.
Decent SEO, but no GEO strategy.
I built a semantic cluster focused on « product entities + repair questions ».
I published comparison pages, stripped-down guides, with FAQ and Product schema markup.
And I connected a call tracker to the sales lines.
Results after 60 days:
37 calls from AI Search.
By day 90, 67 calls.
14 sales at $800 average order value.
$11,200 in revenue directly traced.
Zero clicks.
Zero ads.
Just architecture.
How to trace an AI call without drowning in data
Many e-commerce businesses think it’s too complex.
It’s not.
I break it into 3 parts:
1. A dynamic call tracker.
Place a dynamically generated phone number on each top-funnel page (comparison, product card, FAQ).
CallRail, Ringostat, or WhatConverts do this very well.
2. The UTM parameter « source_medium=organic_ia ».
Configure the tracker to capture the source when the caller comes from a page that appeared in an AI snippet.
It’s manual, but doable.
3. The cross-referenced GEO / calls report.
Wire your tools together: Search Console (AI Overviews filter), Semrush GEO tracking, and the call dashboard.
In 10 minutes, you have a conversion ratio for « AI visibility → call ».
Who’s already winning without you knowing
I audit 15 sites per week.
Those losing organic traffic aren’t the worst-ranked.
They’re simply absent from AI responses.
Conversely, sites I’d never seen before are sweeping up calls because they have a clean Knowledge Graph, well-defined entities, long-tail pages answering ultra-specific questions.
The CallRail study confirms it: the hardest-hit sectors are home equipment, tools, specialized electronics.
Niches where the need for dialogue before purchase is strong.
If you sell a technical product, you’re in the crosshairs.
Not in a year.
Now.
What you should monitor starting tomorrow morning
Don’t look for a magic tool.
Do three things:
- Test your 10 main queries in Google. See if an AI Overview appears, and who’s cited.
- In Search Console, go to Performance, filter « AI Overviews ». Track impressions and click-through rate.
- Install a call tracker, even on 5 pilot pages. In 30 days, you’ll know if the phone is ringing from AI.
This isn’t a gimmick.
It’s the attribution your dashboards are missing.
Audit your site’s AI visibility in 45 minutes
I’ll show you which calls your competitors are already capturing via AI, and deliver the 3 initiatives that’ll make yours ring.
Book a strategic call — 45 minFrequently Asked Questions
Does a standard call tracker automatically capture calls from AI?
No, you need to configure source tracking. Assign a dynamic number to targeted pages and set the campaign to « organic_ia ». CallRail categorizes everything for you.
What’s the difference between classic SEO visibility and AI visibility measurable by phone?
AI visibility often results in zero clicks but directly leads to a call. Classic SEO measures visits; GEO measures inbound calls without a site visit. The data lives in the call tracker, not Analytics.
Do I need a huge budget to appear in AI Overviews?
Not necessarily. I’ve seen sites with a small semantic cluster (30 pages) land calls in 2 months. Budget depends mainly on architecture quality and structured data.
How long before I see AI calls after optimizing?
Plan on 30 to 60 days. Time for Google to digest new structured content. First calls typically arrive between day 45 and day 60.
Which e-commerce sectors benefit most from this shift?
I observe that in many sectors, the buyer needs to speak before purchasing. Technical equipment, professional machinery, spare parts, tools, appliances, bedding, audio gear. Anywhere advice makes the difference.
