AI shoppers went from the worst traffic to the best in a year. Here is the data.
A year ago, a visitor from ChatGPT was worth less than half a visitor from anywhere else. Now they convert better, stay longer and spend more. We went through the numbers, including the ones that cut the other way.
Eighteen months ago, the honest answer to "should a CPG brand care about traffic from ChatGPT?" was "not much yet." The visits were few, and the ones that showed up behaved like tourists. They bounced, they browsed, they rarely bought.
That answer has expired. Adobe's AI Traffic Trends Report for August 2026, built on data covering more than one trillion visits to US retail sites plus a survey of more than 5,000 US consumers, shows AI-referred shoppers now beating other traffic on nearly every quality measure Adobe tracks. This piece walks through those numbers, the year-ago baseline that makes them interesting, and the counterweights that keep them in proportion.
The July 2026 scorecard
Adobe compares visitors who arrive from generative AI tools (ChatGPT, Gemini, Perplexity and similar) with visitors from every other channel. Here is how the two groups compared in July 2026.
Show the data
| Metric | AI vs. non-AI |
|---|---|
| Conversion rate | +60% |
| Time on site | +59% |
| Revenue per visit | +53% |
| Add-to-cart rate | +28% |
| Pages per visit | +22% |
| Engagement rate | +14% |
| Bounce rate | −34% |
Source: Adobe Digital Insights, AI Traffic Trends Report, August 2026
A few details from the report sharpen the picture:
- Conversion was 60% higher for AI visitors, up from 54% in May 2026.
- Time on site was 59% higher, which Adobe calls the largest advantage it has recorded.
- Bounce rates for AI visitors have held between 17% and 20% for more than a year, while non-AI bounce rates sit near 27%.
- Engagement has been 9% to 14% higher for AI visitors since late 2024. It is the most stable number in the set.
Overall volume is moving too. AI-driven traffic to US retail sites was up 62% year over year in July 2026.
The part that makes it a story: a year ago it was the other way round
If AI visitors had always converted well, this would be a footnote. What makes it worth a post is how fast the sign flipped.
Adobe's own wording: twelve months before July 2026, "non-AI visits were worth 128% more than AI visits." Add-to-cart followed the same path. In January 2025, non-AI visitors added to cart at a 69% higher rate. AI visitors pulled ahead for the first time in September 2025, peaked at a 52% advantage in April 2026, and sat at 28% in July.
To put the old figures on the same scale as the new ones, we converted them. If a non-AI visit is worth 128% more, an AI visit is worth 1 divided by 2.28, or about 44% as much, which is 56% less. The same arithmetic turns "non-AI added to cart at a 69% higher rate" into "AI visitors added to cart about 41% less often."
Show the data
| Month | AI vs. non-AI |
|---|---|
| Jul 2025 | −56% |
| Mar 2026 | +37% |
| Jul 2026 | +53% |
Source: Adobe Digital Insights, August 2026. The July 2025 value is our conversion of Adobe's statement that non-AI visits were worth 128% more (1 / 2.28 = 0.44).
Show the data
| Month | AI vs. non-AI |
|---|---|
| Jan 2025 | −41% |
| Apr 2026 (peak) | +52% |
| Jul 2026 | +28% |
Source: Adobe Digital Insights, August 2026. The January 2025 value is our conversion of Adobe's statement that non-AI visitors added to cart at a 69% higher rate (1 / 1.69 = 0.59).
Adobe doesn't claim to know exactly why the flip happened, and neither do we. Two explanations fit the data and probably both play a part. The assistants got better at product questions over 2025 and 2026, so the person who clicks through has already been matched to something close to what they want. And the people using assistants changed: early users were experimenting, while today's users are more often asking a real buying question. The add-to-cart peak in April followed by a drop to 28% in July suggests the advantage will move around as more casual shoppers join.
Who is using AI to shop
Adobe's July 2026 survey found 42% of US consumers had used an AI assistant for online shopping, up from 39% in March. Among people who had, 83% said it improved their shopping experience.
The split by age is wide.
Show the data
| Generation | Have used AI to shop |
|---|---|
| Gen Z | 56% |
| Millennials | 52% |
| Gen X | 38% |
| Boomers | 23% |
Source: Adobe Consumer Survey, July 2026, in the AI Traffic Trends Report
Two other survey results matter for anyone writing product pages:
- 66% of AI shoppers say they visit a brand's website directly to verify what an assistant recommended.
- 89% say they click on links or sources an assistant provides.
That is the bridge between the survey and the traffic numbers. People are asking the assistant first, then coming to check. The person who lands on your product page from ChatGPT has often already decided you are a candidate and wants confirmation. That is a different visitor from someone who clicked a display ad.
Use is also front-loaded. Adobe says 53% of AI shopping activity happens at the top of the funnel (inspiration and discovery), and only 1% after purchase. Older shoppers lean even harder on discovery: Baby Boomers put 64% of their use at the top of the funnel and only 5% near checkout.
How big are the assistants?
The traffic data describes what happens when someone clicks through. The other half is how many people are asking in the first place. The companies report different metrics over different time windows, so the table below is a list, not a ranking.
Google also said in July that it is "sending billions of clicks to websites every week through AI features in Search." Amazon said active users of Alexa for Shopping nearly doubled in the second quarter and interactions were up more than five times year over year.
The counterweights
It would be easy to read the charts above and conclude that every brand should rebuild its marketing around ChatGPT. The same sources argue for patience.
Weekly use is still a minority habit. eMarketer, reporting on Adobe's data, estimates fewer than 17% of US internet users use generative AI for shopping on a weekly basis. "Have ever used" (Adobe's 42%) and "use every week" are very different things.
Plenty of shoppers are not interested. eMarketer also cites research finding 56% of digital shoppers are either uninterested in AI shopping or unaware of it.
Percentages hide the base. A 60% conversion advantage on a small channel is still a small number of orders. Adobe reports AI traffic growth as a percentage and doesn't publish the channel's share of all retail visits in the sections we've cited, so don't assume AI is a large slice of your traffic. Check your own analytics before you move budget. Tip 13 shows how to set up an AI referral channel in GA4.
It's an average across retail. Adobe's figures cover US retail broadly. A grocery brand, a supplement brand and a laptop maker can see very different results, and Adobe's own readability scores (more on that in our grocery piece) show big gaps between categories.
AI traffic is still small, but it's now the best traffic you get. Treat the pages it lands on like your best landing page.
What we'd do with this as a CPG brand
None of this needs a new team. It needs a few changes in priorities.
- Measure it separately. Make a channel group for AI referrers so you can see your own conversion rate for this traffic instead of relying on Adobe's average. If yours is lower than your organic search conversion rate, the landing pages are the first suspect.
- Find the landing pages. In the same report, look at which pages AI visitors land on. For most brands it's a handful of product pages and one or two guides. Those pages are doing the verification job described above, so make sure they answer the obvious questions in plain text: price, size, ingredients, allergens, where to buy.
- Make the facts match. If 66% of shoppers come to your site to check what an assistant said, the worst outcome is that your site contradicts it. Line up your site, Amazon listing and retailer pages on the same facts. Tip 5 has a checklist, and tools like XENA Intelligence's SearchPanel can show you the exact answers assistants give about each product, which makes mismatches easier to spot.
- Write pages that answer one question each. Assistants cite pages that answer a specific question well (tip 6). They also send the visitors who are furthest along in deciding.
- Check the trend quarterly, not daily. The add-to-cart advantage dropped from 52% to 28% in three months. These numbers will keep moving. Look at the direction over a few quarters before you make big bets.
The shift worth noticing isn't that AI traffic is large. It's that the quality question has been answered. A year ago you could reasonably ignore AI visitors because they didn't buy. In Adobe's data, they now buy more often than anyone else.
Questions
Do AI-referred shoppers really convert better than other visitors?
In Adobe's July 2026 retail data, yes: AI-referred visitors converted at a 60% higher rate than non-AI traffic and generated 53% more revenue per visit. A year earlier, non-AI visits were worth 128% more, so this is a recent change.
How many people use AI assistants to shop?
Adobe's July 2026 survey of more than 5,000 US consumers found 42% had used an AI assistant for online shopping, up from 39% in March. Regular use is lower: eMarketer estimates fewer than 17% of US internet users use generative AI for shopping weekly.
Is AI traffic big enough to matter for a small CPG brand?
For most brands it is still a small share of visits. The case for paying attention is quality and direction: the visits convert well, and AI-driven traffic to retail sites grew 62% year over year in July 2026.
Sources
- AI Traffic Trends Report, August 2026. Adobe Digital Insights.
- AI referrals outperform traditional traffic. eMarketer, Aug 19, 2026.
- Alphabet Q2 2026 earnings: CEO remarks. Google.
- Q2 Earnings: Andy Jassy on Amazon Stores growth, delivery speed, and AI shopping. Amazon, Jul 30, 2026.
How we work: every figure in this piece comes from the sources above, which we opened and read before publishing. Anything labeled illustrative is an example, not data.
Related tips
- Can you see traffic from AI assistants in analytics?
- How do you check whether ChatGPT recommends you?
- What content actually gets cited in AI answers?
Keep reading
Your website is becoming a reference book for robots
For twenty years a brand site was a shop window built to pull people in. More and more of its readers are now machines that read it and summarize it for someone else. That changes what a good page looks like.
Stop asking where you rank in ChatGPT
Founders keep asking us what position their brand holds in ChatGPT. The research says that position barely exists. There's a better question, and it can be answered with a spreadsheet and some patience.
Grocery sites are the hardest for AI to read. Here's what to fix first.
Adobe's latest report puts grocery at the bottom of retail for how easily AI systems can read and cite its pages. The gap is widest on homepages and blog content. Here's where the points are lost and how a food brand can win them back.