“ChatGPT sends us the best customers we’ve ever had.” You’ve probably watched a version of that line go by, usually stapled to a screenshot.
And when a screenshot like that lands on your boss’s desk, it stops being a fun fact and becomes a directive: we’re going after AI traffic, and we’re starting now.
I get the excitement. The numbers are a full order of magnitude apart — of course people share them.
But if you’re the one who has to hit the target, you’re already doing the quiet math. Is that 9x one lucky company, or something that’ll happen to us too?
So let’s do two things. First, find where the “9x” actually came from. Then line it up against 973 stores and two hundred million transactions, and see where AI traffic really sits.
Can We Actually Believe The “9x”?
The whole thing traces back to one case study from Seer Interactive, a digital marketing agency, covering a single B2B client from October 2024 to April 2025 (the case study).
They compared conversion rate — the share of visitors who actually sign up or buy — across traffic sources. And the split was loud:
- ChatGPT converted at 15.9%, Google organic search at 1.76%
- The other assistants ran hot too: Perplexity 10.5%, Claude 5%, Gemini 3%
Put ChatGPT next to organic search and you get roughly 9x. Fine — I understand why that traveled.
But read the small print. All those AI conversions add up to 1,370 of them. Organic search, meanwhile, was working off something like 14 million sessions. The rate is huge; the actual number of AI-sourced deals is tiny. It’s a .500 batting average from a rookie who’s had two at-bats.
And this is B2B software — long sales cycle, high price tag. People argue it out with the AI for weeks and click through basically pre-sold. That’s exactly the setup where the gap goes cartoonish.
Where Does AI Rank Once You Zoom Out?
So what happens when you stop staring at one client? Here’s where a bigger study earns its keep — researchers at Goethe University Frankfurt and Mannheim Business School put one out in 2025 (the paper).
The scale is a different animal: 973 e-commerce sites, $20 billion in combined sales, twelve months of first-party data. They stacked 50,000-plus ChatGPT referrals against 164 million transactions from the usual channels.
Translated into something you can actually use, the finding runs:
- AI traffic’s conversion rate and revenue per session beat paid social — the ads you run on Instagram and Facebook
- But they come in under organic search and email
So AI traffic sits in the middle. Better than the people you paid an ad to drag in, worse than the people who came looking for you by name or opened your email. Not a free goldmine, not just hype — a real, mid-tier channel.
The Fun Part: It Depends On What You Sell
Here’s where the study gets genuinely interesting. The AI effect isn’t flat across products — it tracks with complexity.
The more complex the product category, the bigger AI traffic’s share and the more money it brought in. For simple, no-brainer purchases, AI traffic barely pulls ahead of anything else.
Which makes sense the second you say it out loud. Complex purchase, and you sit there talking it through with the AI, letting it compare options and untangle the specs, and you click only once you’re convinced. So you show up ready to buy.
That’s also why Seer’s B2B software case blew the doors off. It landed square on the “complicated product” condition — the exact spot where AI traffic does its best work.
So whether your thing is a think-about-it purchase or a grab-it-off-the-shelf one changes how hard you should bet on this in the first place.
Should We Pour Everything In Right Now?
One more thing worth not missing: how the quality moved over time.
Across the twelve months, conversion rate climbed steadily. Read that alone and you’d sprint. But it’s not that simple, because over the same stretch the average order value — the typical amount per order — slipped.
When conversion goes up and order value goes down, the two cancel. You’re flooring the gas and the brake at once.
The authors’ read is that AI traffic drifts toward the traditional channels over time — but won’t hit parity with organic search inside a year. It’s a young channel, so nobody’s certain. Still, “redirect all our resources today” is a call the evidence just doesn’t support yet.
Conclusion: The Big Number Is Conditional. Measure Yours.
So where does that leave the “9x”? It wasn’t a lie — it was a conditional truth. Measure a long-cycle B2B product while the sample’s still tiny, and the rate can look like anything. Zoom out to 973 stores and AI traffic settles into something calmer: it beats paid social, trails search and email, and earns its money on complex products.
So when the boss waves a screenshot at you, there are really two things to check.
First: is your product the kind people research with an AI before buying? If it’s complex, AI traffic is worth leaning into. If it’s a simple grab, keep the expectations modest.
Second — and this is the one that matters — measure the quality with your own data before you trust anyone’s ratio. Watch conversion and order value together, or you won’t know whether a rising rate is earning more or just getting cancelled out.
The concrete next step is small: in your own analytics, split out AI-sourced visits and set their conversion rate and order value next to your other channels. Someone else’s 9x is a fact about someone else’s conditions. Your answer is sitting in your own numbers — which is the only place it was ever going to be.
Maybe I’m wrong and next year’s data drags AI all the way up to search. But “measure it yourself” costs you an afternoon, and it beats rebuilding the whole budget around a stranger’s screenshot.
Sources
- Kaiser, M. & Schulze, C. (2025), “ChatGPT Referrals to E-Commerce Websites: How Do LLMs Compare Against Traditional Channels?”, SSRN Working Paper, ssrn.com (973 e-commerce sites, $20 billion in combined sales, 12 months of first-party data; 50,000+ ChatGPT referrals vs 164 million transactions from traditional channels. AI referral conversion rate and revenue per session beat paid social but trail organic search and email. The monetary outcome and traffic share rise with product-category complexity. Over the 12 months conversion rate improved while average order value declined; the authors project AI traffic converges toward traditional channels but does not reach parity with organic search within a year.)
- Seer Interactive (2025), “Case Study: 6 Learnings — How Traffic from ChatGPT Converts”, seerinteractive.com (single B2B client, October 2024–April 2025. ChatGPT conversion rate 15.9% vs Google organic 1.76% (~9x); Perplexity 10.5%, Claude 5%, Gemini 3%. AI-sourced conversions totaled 1,370 against roughly 14 million organic sessions.)
