You saw the headline: AI-assisted shopping is forecast to jump 130% this holiday season. Now you are wondering whether a shopper will ask ChatGPT or Google’s AI for a gift idea, get an answer, and never visit your store. You have a few weeks before Black Friday, a shelf of inventory, and no time to rebuild a website. The first thing to know is what the number counts. AI holiday shopping 2026 is forecast to send far more clicks to retail sites, and a click to a retail site is a visit to a store like yours, not a sale that skipped it. The second thing is that you can check how findable your products are in about ten minutes, and most of what you would fix lives in your own store settings, your product feed and your product pages. This post walks through what Adobe measured, how the assistants find products, and the order to fix things in before the rush.
Key Takeaways
Adobe expects AI traffic to U.S. retail sites to rise 130% year over year from November 1 through December 31, 2026, measured as shoppers clicking a link from an AI service to a retail site.
Adobe’s figure does not predict AI-driven sales, and it does not predict that any given shop will be recommended.
Google leans on Merchant Center and Product structured data, ChatGPT uses merchant and product metadata, and Perplexity shows product cards, so showing up in one does not put you in another.
Price, stock, shipping and return details that disagree between your feed, your product page and checkout are the first thing to correct.
Neither Google nor OpenAI promises inclusion, ranking or sales, so measure AI referrals in GA4 and judge the work by what you can see.
What can you do this week so your products can be found?
Start with a ten-minute check. It gives you a baseline, and it will tell you whether you need a small cleanup or something bigger.
| Minutes | What to do | What you are looking for |
|---|---|---|
| 1 to 3 | Ask ChatGPT and Google AI Mode for a gift in your category, with your real price range, recipient and shipping deadline | Whether your brand, your products or a competitor appears |
| 4 to 6 | Open Merchant Center, choose Products, and inspect three important products | Approval status, visibility, price, availability, title, image, and whether free listings are on |
| 7 to 9 | Open GA4 Traffic acquisition and look at Session source and Session medium | Referrals from AI domains, compared with the prior period |
| 10 | Open one product page in a private browser window | Whether the price, stock, shipping, delivery estimate and return policy match Merchant Center |
Write down what you see and the date. Results from an assistant change with the question, the shopper’s location, price, stock and the platform’s own data, so one answer is a snapshot, not a ranking position. Run two or three different gift prompts before you conclude anything. Use the way a real shopper asks: a recipient, an occasion, a budget and a deadline, such as a gift under $75 for someone who loves gardening that can arrive by December 20.
If the check finds nothing wrong, you have your answer for now: keep watching. If it finds a mismatch or a disapproved product, the next sections tell you which fix to make first.
What did Adobe actually forecast, and what does it mean for your shop?
Adobe’s September 28, 2026 release covers U.S. online retail from November 1 through December 31, 2026. Adobe expects AI traffic to U.S. retail sites to rise 130% year over year. It defines that traffic as shoppers clicking a link from an AI-powered chat service or browser to a retail site. The forecast draws on Adobe Analytics data covering more than 1 trillion visits to U.S. retail sites.
| What it measures | What it does not measure |
|---|---|
| Year-over-year growth in clicks from AI chat services and browsers to U.S. retail sites | Sales made through AI assistants |
| A forecast for November 1 through December 31, 2026 | A result that has already happened |
| Traffic across U.S. retail sites in Adobe Analytics | A prediction that your shop will be recommended |
| Shoppers who clicked through to a retailer | The share of all shoppers who use AI |
The percentage also needs its base. Adobe’s January 7, 2026 report found that generative-AI referrals to retail sites rose 693.4% during the 2025 holiday season compared with 2024, and Adobe said the user base remained modest. A large percentage on a small starting number is still growth, and it is worth preparing for, but it is not a signal that stores are being bypassed. Because Adobe counts the click to the retailer, its number describes visits arriving at shops. What decides whether some of those visits arrive at yours is how well your products are described, priced and stocked where the assistants look.
Adobe also keeps a holiday shopping report page with its ongoing figures, which is the place to check for updates rather than a headline summary.
How do ChatGPT, Google and Perplexity find products, so you know where to look?
There is no single shared product database. Each assistant draws on its own mix of feeds, structured data and crawled pages, so appearing in one does not establish that you appear in another.
- Google (AI Mode, AI Overviews and Gemini shopping): These use Google’s Shopping Graph and Merchant Center product data. Google recommends providing product data through both Merchant Center feeds and Product structured data, and says using both helps it correctly understand and verify the data. Google’s Merchant Center guidance on AI performance recommends high-quality, current product data with complete attributes.
- ChatGPT: OpenAI’s shopping documentation says results use product and merchant metadata from first-party and third-party providers, public retail pages, prices, availability and reviews. Shopify catalog data is integrated, and other merchants can apply for direct product-feed access. OpenAI says the results are independently selected and are not ads.
- Perplexity: It shows product cards based on relevance, authority, product detail, reviews, price, availability and specifications. Its Instant Buy program supports checkout for eligible merchants and products.
OpenAI also notes that shopping results can contain errors and suggests visiting the merchant’s site for current details. That is a reason shoppers still click through, and a reason your own page needs to be the accurate, complete version of your product.
Some purchases can finish inside an assistant, and that is a separate question from this forecast. If you are deciding whether to let AI agents buy on a shopper’s behalf, our earlier post on AI shopping agents and your online store covers that decision.
Which fixes get your store ready fastest, in what order?
Work down this list. It follows the order platform documentation treats as most important: get the product data right and approved, keep it consistent, then make the buying details clear. Time estimates are typical for a small catalog.
| Problem | What to do | Typical time |
|---|---|---|
| Incomplete product data | Complete the required feed attributes and product markup | A few hours for a small catalog |
| Disapproved products | Read the Needs attention details in Merchant Center, correct the data or policy issue, resubmit | Hours to several days |
| Price or stock differs between feed, page and checkout | Connect your store platform to a reliable feed or update it more often | Hours to several days |
| Unclear shipping or returns | Publish clear policies and map them to Merchant Center and structured data | One to four hours |
| Thin titles and descriptions | Rewrite priority products around factual attributes such as material, size, recipient and delivery time | One to three hours per product group |
| Missing analytics attribution | Preserve referrers, standardize UTM values on links you control, test your reports | One to four hours |
Start with your best sellers and your likeliest gift items, not the whole catalog. A few products with complete, matching data are better than a catalog full of gaps.
Each step in practice:
- Merchant Center and free listings. Create or verify your Merchant Center account, submit the catalog and opt into free listings. Then check that each important product has an accurate title, description, product URL, image, price, availability, brand, condition and identifier where one applies.
- Product structured data. Add Product and merchant listing markup to product pages, following Google’s Product documentation. The vocabulary comes from schema.org’s Product type, and shipping details use OfferShippingDetails.
- One version of the truth. Keep feed data, page data, checkout price and inventory synchronized. A feed that says in stock while checkout says sold out is the kind of mismatch that gets products disapproved and shoppers annoyed.
- Shipping and returns. State shipping costs, delivery timing, your return policy and how to reach customer service. Holiday shoppers ask for a delivery date first.
- Reviews and variants. Add genuine product reviews and make product variants explicit, such as size and color as separate, clear options.
- OpenAI feed access. If you sell on Shopify, your catalog may already be integrated. If not, OpenAI’s documentation describes applying for product-feed access.
If your store runs on a custom build, some of these steps, like a live feed or accurate structured data, may need a developer. That is the kind of work custom e-commerce development covers.
How do you tell whether AI is already sending you visitors?
You can see it in GA4 without any new tools. Open Reports, then Acquisition, then Traffic acquisition. Look at Session source, Session medium, landing page, sessions, engagement, conversions and revenue. Look for referrals from domains such as ChatGPT, Perplexity, Gemini or Copilot.
Two limits matter. GA4 classifies a visit as direct when it has no identifiable referral information, so AI-sent visits can be undercounted or grouped with direct traffic, especially when a platform opens a browser without passing a referrer. Read how GA4 handles direct traffic before you decide a low number means low interest.
UTM parameters help on links you control, such as email, creator campaigns and paid promotions. Use consistent values such as `utm_source=chatgpt`, `utm_medium=referral` and `utm_campaign=holiday_2026`. You cannot add UTMs to links an AI platform generates on its own. Our post on UTM parameters for a small business covers naming them consistently.
Compare the same date range with the prior period, and watch direct traffic to product pages alongside AI referrals. Together they give a fair reading of a channel that is measured imperfectly.
What can you stop worrying about before Black Friday?
Some worries cost time and do not change your data.
- A special chatbot persona or dozens of generic AI articles. The work that counts is product data, not new content about AI.
- Renaming every product around the phrase AI shopping. Google’s guidance recommends complete attributes and current data. Where shoppers use natural phrasing, such as a recipient, budget or deadline, state those factual details in titles and descriptions without stuffing phrases in.
- Rebuilding the store before the holidays. A cleanup of the feed, the markup and the shipping and returns pages is a smaller job.
- Treating every AI answer as a ranking. Results are contextual and can change with the query, location, inventory, price and platform data.
- Assuming a mention equals a sale. Adobe counts clicks to retail sites, while some platforms can complete checkout without a site visit.
A shopper still has reasons to reach your page: to verify the current price and stock, look at images and variants, read the return policy, confirm delivery timing, judge whether they trust the brand, or use a discount code. The more useful your product page is on those points, the more it is worth after an assistant sends someone to it.
When is it worth paying for help, and when should you do this yourself?
You can run the ten-minute check alone, and you can do the first several fixes in your store’s admin. Outside help earns its cost when the check turns up a specific problem that is slow to diagnose during a busy season: a feed that will not sync, products that keep getting disapproved, markup that does not validate, pages that are hard to crawl or render, analytics that lose the referrer, or checkout data that does not match the page.
If the catalog is already accurate and visible, the right move may be to keep monitoring and spend nothing. Nobody can buy a guaranteed recommendation from an assistant, and neither Google nor OpenAI promises inclusion, ranking, recommendation, traffic or sales from feeds or markup. Before you buy a broad package, ask for a look at your data first. A capable provider should be able to show you the affected products, the platform requirement involved, the exact correction, how it will be verified and what the result cannot promise.
Check what you know about the holiday forecast
Pick an answer to begin.
1. What does Adobe’s 130% figure measure?
2. Which two sources of product data does Google recommend using together?
3. What should you fix first when the ten-minute check finds a problem?
Frequently Asked Questions About ai holiday shopping 2026
What did Adobe forecast for AI shopping in 2026?
Adobe forecast a 130% year-over-year increase in AI traffic to U.S. retail sites from November 1 through December 31, 2026, measured as shoppers clicking a link from an AI-powered chat service or browser to a retail site.
Is 130% a sales forecast?
No. It measures clicks from AI services to retail sites. It does not predict AI-driven sales, and it does not predict that any specific shop will be recommended.
What actually happened in the 2025 holiday season?
Adobe reported that generative-AI referrals to retail sites rose 693.4% compared with 2024 during November 1 through December 31, 2025, and said the user base remained modest.
How can a shop appear in Google’s AI shopping results?
Use Merchant Center with free listings turned on, keep product data accurate and complete, and add Product structured data to your product pages. Google does not promise placement.
How can a shop appear in ChatGPT shopping results?
Keep public product pages accurate and complete. Shopify catalog data is integrated, and other merchants can apply for OpenAI product-feed access. OpenAI does not promise inclusion or ranking.
How can I see AI traffic in my analytics?
In GA4, open Traffic acquisition and review Session source and Session medium for AI domains. Some AI visits may be grouped as direct, so also watch direct traffic to product pages.
The Bottom Line
Adobe’s forecast is about traffic arriving at retail sites, not sales, and it is not a reason to think shoppers will skip stores. It is a reason to spend an afternoon making sure your products are described the same way everywhere a shopper or an assistant might look. Run the ten-minute check, correct any price, stock, shipping or returns mismatch, get your best sellers approved and marked up, and set up GA4 so you can see what arrives.
After that, watch the results for a few weeks. Some AI visits will show as referrals, some will land in direct traffic, and some answers will change from day to day. That is normal for a channel this new, and it is why steady product data beats any one-time trick. If a step reveals something you cannot fix from your admin, that is the time to bring in help.
If the check turns up a feed, markup or checkout problem you would rather not chase during the holidays, we can help. At Web Leveling we work on generative engine optimization and store development, and we will tell you plainly when your own settings are enough. Send us the results through our contact page and we will look at the specific products and data involved. We work with small and medium businesses across the country and overseas.
Terms
Terms used in this post
Tap a term to see what it means.
AI-assisted shopping traffic. In Adobe’s forecast, visits that begin when a shopper clicks a link from an AI-powered chat service or browser to a retail site.
Merchant Center. Google’s tool for submitting product data so products can appear in Google shopping surfaces, including free listings.
Free listings. Product listings in Google that do not require paid advertising, turned on inside Merchant Center.
Product structured data. Markup on a product page that describes the product’s price, availability and other details in a form search systems can read.
Product feed. A file or connection that sends your catalog data, such as titles, prices and stock, to a platform.
Referral traffic. Visits that arrive from a link on another site or service, which analytics tools report by source and medium.
UTM parameters. Tags added to a link you control so analytics can report the source, medium and campaign of the visit.
