Should We Be Moving Budget To ChatGPT Yet?
For about a year now, I haven’t gone a week without an ecommerce person telling me their ChatGPT traffic is climbing. And the raw numbers back them up: Adobe clocked GenAI-driven US retail traffic up 1,200% against July 2024, and up 1,950% year-over-year on Cyber Monday.
Stare at that and the obvious thought shows up. Is this the new main channel? Should the budget follow?
I get the pull. I’ve felt it too. But the thing a marketer actually cares about isn’t traffic volume — it’s whether that traffic turns into sales.
So let’s look at the quality side, not just the size.
What 973 Stores Say About ChatGPT’s Conversion Rate
Kaiser and Schulze, at Goethe University Frankfurt and Mannheim Business School, tracked 973 ecommerce sites for twelve months — roughly $20 billion in combined revenue — using to line ChatGPT referrals up against the traditional channels.
Here’s where ChatGPT landed: its conversion rate beat paid social and trailed organic search.
That’s worth sitting with. It’s not a weak channel. Without spending a cent on ads, the buying behavior after the click already clears a real bar. So the first conclusion is small but honest — it’s worth trying.
The same study adds one catch, though. The more complex the product category, the better the LLM (large language model) referrals performed. So this doesn’t pay off evenly across everything you sell, and that changes how you should read your own numbers.
The Quality Is High. The Volume Is Still Tiny.
Alhena AI’s comparison of 329 brands across 9 channels tells the same story from another angle. LLM traffic converted at 2.47% — 4th of the nine channels, and well ahead of paid social at 0.52%. Engagement ranked 2nd. Quarter-over-quarter growth ran 40%.
But traffic share was about 1% of the total. The visitors are good; there just aren’t many of them yet.
Skip past that and jump straight to “the CVR (conversion rate) is high, let’s move the budget”, and you’ll probably misread the whole thing.
Adobe’s data draws the same shape. LLM traffic browses deeper and bounces less, and yet its conversion rate still runs 9% below the traditional average. So people are getting further into the decision — they’re just not closing at the same rate yet. It’s a “consideration is working, the last step isn’t quite there” phase.
Why One Report Says “High” And Another Says “Low”
So you’ve got “CVR is high” and “CVR is still low” sitting side by side. That’s not a contradiction. They’re measuring different crowds.
- Alhena compares channels within the LLM world.
- Adobe compares LLM traffic head-to-head against traditional traffic.
- Kaiser and Schulze compare across channels over a long window.
Three questions, three answers. And the practical read is simple: don’t look at CVR on its own. Put CVR and referral share side by side over the same window. Those two axes together catch the moves a single number hides — “quality’s improving but volume isn’t keeping up”, or “volume jumped but the money got less efficient”.
The other move is to split by category. Judge a complex, deliberated purchase and an everyday staple in the same bucket and the LLM win blurs. Separate them and the investment call gets a lot clearer.
Conclusion: Lock In Two-Axis Measurement Before You Move Budget
ChatGPT traffic is already producing buying behavior you can’t wave off. It’s a strong-CVR channel in its early innings — which is exactly the kind of thing you want to measure well before you bet on it, not after.
At a minimum, split chatgpt.com and the other AI referrers out of your “direct” bucket, and put referral share and category-level conversion on the same dashboard. Then read them together, week over week. When your own curve sits next to the Kaiser & Schulze 973-store benchmark and Adobe’s tracking, you get a much steadier sense of what to expect — and you’ll spot the moment the volume side starts catching up to the quality side.
Sources
- Kaiser, M. & Schulze, C. (Goethe University Frankfurt / Mannheim Business School), “ChatGPT Referrals to E-Commerce Websites: How Do LLMs Compare Against Traditional Channels?”, SSRN Working Paper, 2025, DOI: 10.2139/ssrn.5585812
- Alhena AI, “329 Brands, 9 Channels: The AI Commerce Report for 2026”, alhena.ai, 2026
- Adobe, “Adobe Analytics: Traffic to U.S. Retail Websites from Generative AI Sources Jumps 1,200 Percent”, Adobe Blog, 2025
