How Much Traffic AI Assistants Actually Send Your eCom Store (Spoiler, It's 0.20%)

There's a study that made the rounds this summer and it's worth two minutes of your time. A guy pulled two years of analytics across 10 established Shopify stores to answer one single question, how much traffic AI assistants like ChatGPT actually send. His first answer in July was 0.10%. He got torn apart in the comments, everyone told him he was undercounting. So he re-pulled everything through August 31st. The result for August alone, 0.20% of revenue.

0.20%. That's where we really are once we stop fantasizing.

Let me tell you where I'm coming from. At OMOSA I spend my days inside GA4 and Shopify back-offices for fashion and lifestyle brands, cross-referencing traffic sources, conversion rates by collection, net sales by country. So when I see these numbers, I don't read them as a tech headline. I read them like I'd read a client account. And what I'm about to tell you isn't to reassure you, it's to get you to stop wasting time on the wrong problem.

The numbers, no filter

Let's go through what the study shows, figure by figure.

In August, ChatGPT sent 6,868 visits to these 10 stores. That's a record for them, and it's 116x more than the 59 visits in August 2024. 59 visits. Across 10 established stores. A year ago.

Overall AI traffic is up 2.3x versus last summer. Revenue attributed to AI, on the other hand, is up 5.3x, going from about $10,000 last summer to $56,000 this summer. So revenue is climbing more than twice as fast as traffic. That's the thing that really caught my eye.

Over the full year, AI accounts for $131,000 out of $109.6 million in revenue. Its share doubled in two months. And it's still tiny.

One detail I really like in the methodology, he kept each original figure on its original window so we could compare, and when he re-pulled the full history, it matched the July pull to within one session out of 94 million. Clean, verifiable work.

Why it still matters

I'm not going to sell you on 0.20% being a revolution. It's not a revolution. But there are two things in here worth noticing.

The first is the speed. Going from 59 visits to nearly 7,000 in one year on a single channel is not nothing. In my line of work, when I see a source doing 116x year over year, I put it on watch. Not because it carries weight today, but because the slope is telling me something. Usually, a channel behaving like this either plateaus fast, or becomes a real topic in 18 months.

The second, and this is the most important one, revenue is climbing twice as fast as traffic. Translation, people coming in via ChatGPT convert better and spend more than average. That makes sense. Someone who asks an assistant "find me a wool coat under 300 euros" and clicks through is much further down the funnel than the guy scrolling Instagram at 11pm. These people aren't coming to browse, they arrive with a purchase intent already formed.

It's exactly like a physical store. Traffic on your site is like foot traffic in a shop, not everyone has the same intent. The customer who walks in saying "I'm looking for this exact model" is worth ten times the window-shopper taking shelter from the rain. Right now, AI traffic looks like that first customer, the one hunting for the specific model.

What I'd do in your shoes (and what I wouldn't)

Let's be clear on priority order, because this is where people screw up.

If, say, you do 0.20% of revenue through AI and your conversion rate has dropped 15% over the last three months, you don't work on AI. You work on your conversion. There's no comparison. I spent an evening recently pulling organic and direct traffic from an account just to understand why conversion was tanking, and the answer was in search, not in some passing fad. The money is in your fundamentals. Not in a channel at 0.20%.

What I would still do, and it costs almost nothing.

First thing, measure it properly. Most of the brands I work with can't even isolate AI traffic in GA4. Build a segment, tag referrals like chat.openai.com, perplexity and the rest, and look at your sessions, your add-to-cart rate, your conversion rate on that specific segment. Even if it's 200 sessions a month, you want the data. You can't steer what you don't measure.

Next, work on your product page and your content. AI assistants recommend what they understand. If your product page has a clear title, a precise description, clean specs, a readable price, you've got a better shot at showing up. Just last week, a product page with no specs and no decent visual was dragging down a client's conversion. A machine reads your page like a human, only worse. So sloppy product content closes two doors instead of one.

Lastly, look at where you send that traffic. It's a reflex I apply to every channel. Traffic diluted across 40 collections and 300 products converts worse than traffic concentrated on your best-sellers. If AI is sending you people, make sure the pages that surface are the ones that convert, not some catch-all collection sitting at 1.3% conversion.

What I wouldn't do is rebuild my entire content strategy "for AI," pay a vendor who promises to "rank you in ChatGPT," or move budget away from channels that bring in 95% of your revenue. We're not chasing a revolution here. We're just doing things right. And doing things right here means taking care of your fundamentals first. If your budget and your margin feel fuzzy, start there, build your ad budget off your margin, not your revenue, before chasing an emerging channel.

The thing that actually costs you is somewhere else

The real danger is elsewhere. You let yourself get distracted. There's always a new shiny channel showing up, and every time the same panic kicks in, "we need to be on it." I've watched brands burn weeks on a 0.20%-of-revenue topic while their add-to-cart rate dropped 15% year over year and nobody was looking.

I always reason the same way. I start from the objective, then work back to the real problems, and only then decide what to do. If your objective is growing profitable revenue, the number-one problem in 2025 is rarely AI. It's your conversion. Or your retention. Or your contribution margin, depending on where things are breaking in your business. AI is a line to watch on your dashboard, not a project.

That said, the slope is worth a glance. 116x in a year, revenue climbing twice as fast as traffic, share doubling every two months. Keep it on your radar, measure it, take care of your product pages. But keep your hands on the wheel.

Where to actually start

Three actions, in order.

1. Isolate AI traffic in your GA4 and let it run for a month. You'll know where you really stand, in your own business, not across 10 American stores.
2. Audit your 20 most important product pages, title, description, price, specs. If a human struggles to understand it, so does a machine.
3. Check that your incoming traffic lands on your best-converting pages, not in a catch-all.

And if before all that you want to know where your money actually goes, an audit that starts with your P&L and not your campaigns will tell you more about your growth than any study on ChatGPT. It's less sexy. But that's what pays.