Can ChatGPT shoppers find your store?
A growing share of holiday shoppers will not start on Google this year. They will ask an assistant what to buy, and the assistant will open a few stores, read them, and recommend one. If yours is a set of pictures on a page that changes address every season, it will not be the one. Here is what an assistant can read, what it cannot, and what a small seller can fix before October on JustCopy or anywhere else.
Where the shoppers are coming from
Adobe Digital Insights tracks visits to U.S. retail sites that arrive from generative AI tools, the way a click from a search result is tracked. Two years ago the number was too small to report. It is not small now, and the visitors it describes are not window shoppers. The figures below are from Adobe's reports and the coverage of them, checked in September 2026.
693%
Year-over-year growth in AI-sourced traffic to U.S. retail sites in the 2025 holiday season, November through December (Adobe Digital Insights, checked September 2026, as reported by TechCrunch).
393%
Year-over-year growth in the first three months of 2026 (Adobe Digital Insights, checked September 2026, as reported by TechCrunch).
54% better
How AI-sourced visitors converted compared with visitors from other sources in May 2026, a year after they converted at nearly half the rate of everyone else (Adobe Digital Insights, checked September 2026, as reported by Digital Commerce 360).
37% higher
Revenue per visit from AI-sourced traffic compared with non-AI traffic in March 2026 (Adobe Digital Insights, checked September 2026, as reported by TechCrunch).
The conversion figure is the one to sit with. In March 2025, a visitor sent by an AI tool bought less often than anyone else. A year later the same kind of visitor bought more often than anyone else. The shopper who asks an assistant has already narrowed the choice, and the store that gets the visit is the one the assistant could read well enough to recommend.
Adobe's June 2026 report added the other half: even in the retail categories that do best, 30 to 40 percent of the content on retail pages is overlooked or not captured by AI tools (Adobe Digital Insights, checked September 2026, as reported by Digital Commerce 360). The traffic is growing faster than the pages are being fixed. For a small store that is an opening, because the fixes are about what is on the page and how it is written.
What an assistant can read
An assistant that shops for someone does what a hurried person does: opens the page, skims the headings, looks for the product and the price, checks how to buy, and moves on. The difference is that it reads the HTML the server sends, not the picture on your screen. Seven things decide whether it finds what it needs.
Text the server sends
An assistant fetches the page and reads what comes back. If the product name, the price and the description are in that first response, it has them. If they arrive later through a script, it may see an empty shell.
A headings outline
One H1 that says what the store is, then H2s for the sections: products, shipping, hours, contact. An assistant uses the outline the way a person uses the table of contents.
Prices written as text
"Beeswax candle, $18" in a paragraph is a fact. The same words in a photo of a price tag are not. Write the currency and the number next to the product name.
Hours and an address
For a store with a counter, the hours and the street address decide whether the assistant suggests you to someone nearby. Put them in text, once.
A checkout that works
A cart that adds the item and a checkout that finishes without an error is the difference between a mention and a sale.
One stable URL
One address for the business, the same one everywhere you are listed. Three addresses split whatever trust the store earns three ways.
Structured data
Schema markup that names the business, the products and the prices in a form built for machines. It confirms what the text already says. It does not replace it.
What it cannot read
Most stores that are invisible to assistants are invisible for one of these four reasons, and none of them is about the size of the business.
Words inside images
A banner that says "Free shipping over $50" as a picture is invisible. So is a photographed menu or a price list exported as a JPEG.
Content that needs a click first
Prices behind a "View details" button, sizes in a dropdown that renders after a script runs, a catalog that loads as you scroll. Some assistants run scripts, many do not, and none wait long.
Only a social profile
An Instagram grid is photos with captions on a page the assistant may not be allowed to open. It cannot find your prices, your hours or a way to pay there.
A page that changes address
A new link every season, a builder subdomain that switched last month, a redirect that fails: each one breaks the chain between the mention and the sale.
There is a simple test. Select everything on your page, copy it, and paste it into a plain text editor. What you see there is close to what an assistant sees. If the paste is a few words and a lot of blank lines, the page is images and scripts, and the work is to get the words into the page.
How a JustCopy store looks to an assistant
JustCopy pages pass that test by default. Every generated page is server-rendered HTML: the product names, the prices, the hours and the address are in the first response, before any script runs. There is nothing to configure. The words on the screen are the words in the page.
Every published app also gets one public page at justcopy.ai/apps/ followed by its id, with its own title, description, an H1 with the name of the business, WebApplication structured data, and the app's own headings written out as text. An assistant reading it finds the outline of the store without opening the app itself, at one stable address that stays the same through every change you make.
The instant online-store app is the fastest way to get there. Describe the store in one sentence and the page is on screen in seconds with a catalog, a cart and a simple checkout. Then change the words: on a platform-built page, a text, price or section change is free and takes seconds. Type the real product names and the real prices, and the page is yours.
Publishing is one tap on every plan. On the Free plan you get 3 AI builds a day and publish with the JustCopy badge. On Hobby, $9.99 a month, the store is live on your own domain with no badge, and the domain is the address you give to every profile and listing. The plans are on the pricing page.
Before October: the checklist
Six items, none of which needs a developer. Do them in order, because the first three fix most of what an assistant misses.
- 1
One address
Decide the single URL for your store and use it in every profile, bio and listing. If you have several, redirect the others to it.
- 2
Names and prices as text
Run the copy-and-paste test above. If a product or a price is missing from what you pasted, it is missing for the assistant too.
- 3
A clear outline
One H1 with the store name and what it sells. An H2 for products, shipping, hours and contact. No section that only exists as a photo.
- 4
Hours and address in text
Write them once in the contact section, in plain words, and keep them current through the holidays.
- 5
Buy something yourself
Add an item, go to checkout, finish the order on a phone. Fix anything that stops you before a shopper meets it.
- 6
Structured data
Check that the page carries schema for the business and the products, then move on. It is the last item on this list for a reason.
The holiday traffic Adobe measured went to the stores an assistant could open, read, and trust enough to send someone to. That is a low bar, and most small stores are still under it. Clear it in September and the shopper who asks in November finds you.
Build a store an assistant can read. Free, in seconds.
One sentence becomes a store with a catalog, a cart and a checkout, as plain text on one stable address. Publish with one tap.
Questions
Do AI assistants really send shoppers to small stores?
They send shoppers to stores they can read. Adobe Digital Insights measured AI-sourced traffic to U.S. retail sites up 693 percent year over year in the 2025 holiday season and up 393 percent in the first quarter of 2026 (checked September 2026). The assistant does not care how big the store is, only whether the product name, the price and a way to buy are plain text on a page it can open.
What does an AI assistant see when it opens my page?
Mostly the HTML the server sends. Text in headings and paragraphs, prices written as text, the address and hours, and structured data are easy. Words inside images, menus that only appear after a script runs, and prices that load later are often missed or read wrong.
Do I need structured data?
It helps, but plain text comes first. A page whose product names and prices are visible text with a clear headings outline is readable with or without schema. Add structured data on top, not as a patch for a page of images.
Is a JustCopy store readable by assistants?
Yes. Every JustCopy page is server-rendered HTML, so the words are in the first response. Every published app also has a public page at justcopy.ai/apps/ followed by its id, with its own title, description, H1, WebApplication schema and the app headings written out as text.
How long does it take to set up a store this way?
The instant online-store app is built from one sentence in seconds, with a catalog, a cart and a simple checkout. Text, price and section changes on a platform-built page are free and take seconds. Publishing is one tap on every plan.
What does it cost to publish?
The Free plan gives you 3 AI builds a day and publishes with the JustCopy badge. Hobby is $9.99 a month for your site on your own domain with no badge. The full list is on the pricing page.