In its October 29, 2026 report, Adobe found that AI-referred visits to U.S. retail sites generated 83% more revenue per visit than non-AI visits in September. Their visit-to-purchase conversion performance was 90% higher.
For a retailer, those figures make AI shopping worth investigating. Start by finding which products your own AI-referred visitors buy and what those orders leave after costs. That gives you something concrete to test with a paid budget.
Before increasing spend, work out what you can afford to pay for another customer. Adobe gives you a reason to investigate; your store’s costs determine the budget.
What the Report Tells You About AI Shopping
Adobe measures this traffic through people clicking links to retail sites from AI experiences. Its broader analysis covers more than a trillion U.S. retail visits, 100 million SKUs and 18 categories. That is the scope of the retail analysis, not a trillion AI-referred visits.
These are observed referrals, rather than results for a specific paid advertising product. They also miss some journeys, such as a shopper reading an AI answer and buying later through another route. Keep that scope in mind when comparing the report with your own store.
Higher revenue per visit can come from more people buying, larger orders or both. For the same period and records:
Revenue per visit = orders per visit × revenue per order.
The 83% figure therefore does not mean baskets were 83% bigger or profits 83% higher. Acquisition and fulfillment costs still need to be paid. The figures describe this U.S. September comparison; percentages from other months answer a different question.
Find the Commercial Opportunity in Your Own Store
If you can identify AI-referred visits, open the product and order records behind the average. Which products did people buy? Were they new customers or existing buyers? Did they keep the orders?
Look for a product-level clue. If AI-referred visitors keep choosing a commuter bag with clear laptop measurements and dependable stock, that gives a bag retailer a concrete offer to investigate. A store-wide average is harder to act on.
Put the number of visits and orders next to the ratios. A few purchases can change a small group’s apparent performance sharply. Compare similar periods and, where possible, similar customers and products. Support visits and promotional shopping visits need not behave alike.
You can still consider a small paid test when referral data is thin. Choose a product with a clear buying need, prepare the page and set a limit the business can afford. The purpose is to learn whether that offer can acquire customers at a workable cost.
Translate Customer Value Into an Affordable Test
Once you have a candidate product, work back from the money its orders leave after product, fulfillment, payment and expected return costs. Keep enough toward overheads and profit before setting an acquisition allowance.
For a $40 item, suppose those costs leave $8 per order the customer keeps, before acquisition. Keeping $5 for the business leaves $3 to win the order. That is the number to bring into the budget discussion, even when the channel’s revenue per visit looks impressive.
Match the allowance to the action on the bill. A $3 limit per retained order cannot be used as a $3 click budget when most visits may not buy. For sales-based billing, account for returns and the campaign’s actual charge adjustments before setting the target.
Use the First Test to Answer One Business Question
For the bag retailer, the question could be: can a paid AI route acquire suitable buyers for the in-stock commuter bag at an acceptable cost? The destination needs clear fit information, realistic delivery terms and a working purchase path.
With PingPlus, the retailer can explore a paid campaign for that offer in the markets it serves. Confirm the available AI delivery options, recorded action and spending limit before launch. Use the campaign to find out what suitable buyers cost to acquire and what their orders leave the store.
Follow the orders through fulfillment and returns before raising that limit. A broken purchase path needs repair. Suitable customers at an affordable cost may justify continuing. Orders that cannot cover acquisition costs are a reason to change the offer or stop.
The paid acquisition and measurement guide helps distinguish this operating result from a stronger claim about sales caused by advertising. The latter needs a credible comparison or experiment.
Adobe gives retailers a reason to pay attention. Your margins and the orders you can actually serve decide what that attention is worth funding.
FAQ
Is revenue per visit the same as ROAS?
No. Revenue per visit divides revenue by visits; ROAS divides attributed revenue by advertising spend. Neither deducts every cost or measures profit.
Does Adobe’s report prove ChatGPT Ads will be profitable for my store?
No. It reports observed referral performance for a market and period. It does not establish your paid acquisition cost, product contribution or the sales caused by a particular campaign.
What if I cannot identify enough AI referrals in my data?
You can still consider a small paid test. Start with an offer you can fulfill, a spending limit and a way to check the orders. A handful of organic referrals is too little to predict the test’s conversion rate.
When do I need an incrementality study?
When you need to know how much business would not have happened without the advertising. That calls for a credible comparison and enough data. An attribution report tells you which activity received credit, which is a different question.




