How To Get Your Products Recommended by AI in the 2026 Holiday Shopping Season

Shoppers now look for gifts inside AI tools like ChatGPT, Google AI Mode, Gemini and Claude. Adobe recorded a 693% year-over-year rise in generative AI traffic to US retail sites during the 2025 holiday season, and forecasts a further 130% rise for November and December 2026, with 141% on Thanksgiving Day (Adobe, Adobe 2026 forecast). Personal AI agents add a second channel this year: Meta's Muse and Instinct search for products, track prices and complete purchases with the shopper's approval.

A typical holiday request reads: "a beginner espresso machine under $300 for my sister who drinks lattes, delivered before December 20." An AI system answers it by matching price, features, delivery dates and reviews against the request. A product with missing or conflicting data on any of those points is easy to leave out.

AI shopping optimization is the practice of making product data easy for AI systems to retrieve, interpret, verify and act on, so a product can be matched to a shopper's request. It applies to answer engines such as ChatGPT, Gemini and Google AI Mode, and to personal agents such as Meta's Muse that shop on a person's behalf. The table below lists the main AI shopping tools, where each one finds products, and what a brand controls on each.

1. Put your full catalog into the product feeds AI tools read

The share of ChatGPT Shopping recommendations drawn from merchant feeds rose from 8% to 62% on July 10, 2026, and reached about 65% by September 3. Over the same period the top 10 stores' share of recommendations rose from 22.5% to 41.8%, so products outside the feeds compete for fewer slots (Search Engine Journal).

• Shopify stores: In Shopify admin, go to Sales channels > Agentic. Confirm AI channel access is enabled for ChatGPT and Microsoft Copilot, and all holiday SKUs are eligible for Shopify Catalog. For Google AI Mode, verify product sync through the Google & YouTube channel.
• Other platforms: Request direct product feed access from OpenAI, and ask your commerce platform or feed provider whether it supports OpenAI's product feed specification.
• Google and Meta: In Merchant Center, populate product_highlight, product_detail and relevant related_product (use part_of_set for related gift-set products). Keep your Meta Commerce Manager catalog complete and synced for Meta AI shopping experiences.

In ChatGPT, the product cards now draw mostly on merchant feeds, while the written answer still cites publishers, retailers and forums.

2. Add a comparison table and a gift details block to each priority PDP

Holiday shoppers ask AI tools to compare a product with similar ones and to filter by budget, recipient and use. An AI engine can only match those details when the PDP states them in text. Two additions to the PDP cover most gift requests.

• Comparison table: Compare the product with its nearest siblings in 4 to 6 rows: price, the specs that differ, and a "best for" line. For example: "E-20, $279, built-in grinder and steam wand, best for latte drinkers." Clarify what the extra money buys.
• Gift details block: State who the product suits (age range, skill level, use case), what's in the box, size or compatibility, and whether gift wrap and gift receipts are available.
• Gift sets and bundles: List every item by its full product name and size, link each item to its own PDP, and use the same set name in the feed title, the PDP H1 and Product schema.

3. Spell out every holiday deal in text on PDPs

AI engines can only recommend a holiday deal they can read. Put each deal's details (product, sale price, previous price, dates and who qualifies) in text on the PDP and in your product feed.

• On the PDP, in text: the exact model and variant, the sale price and previous price, start and end dates with time and time zone, who qualifies ("no code needed", "members only"), and the shipping total. A banner reading "Up to 40% off select styles" doesn't give an AI tool any of these facts.
• In the feed and schema: Set sale_price and sale_price_effective_date in Merchant Center, and keep Offer price and priceValidUntil in Product schema identical to the visible price. Discounts applied only in the cart or by code are invisible to AI tools, so show the discounted price on the PDP.
• Delivery and returns: Publish order-by dates per shipping method ("Order by Dec 18 for standard delivery by Dec 24") and the holiday return window. Mirror both in Merchant Center shipping and return settings and in OfferShippingDetails and MerchantReturnPolicy markup.

An AI tool can recommend a holiday offer only when the page states the model, the price, the previous price, the dates and who qualifies.

4. Get every review from your review widget into the PDP's HTML

41% of ChatGPT's searches for product questions look for reviews (Profound). Third-party review widgets usually load reviews with JavaScript, often with only the first few in the page HTML, and most AI crawlers do not run JavaScript, so a PDP with 900 reviews can look to them like a page with five.

• Check: Load the PDP with JavaScript disabled (or fetch it with curl) and search the source for a sentence from your 20th review. If the sentence is missing, AI crawlers cannot read that review.
• Fix: Ask your review vendor to switch on server-side rendering or its SEO HTML option. If it has neither, pull reviews through the vendor's API, render them into the page on the server, and link further review pages with plain <a href> links (?page=2) instead of a "Load more" button.
• Mark up: Add AggregateRating and Review schema that match the review count and reviews visible on the page.

5. Pitch the publishers AI cites for holiday shopping in your category

89% of the Black Friday pages cited in AI answers came from publishers (Profound). In a given category, those pages are a short list, and many publishers refresh the same gift guide and deals URLs every year.

6. Test whether a shopping agent can buy from your site

Personal agents such as Meta's Muse (launched September 8, 2026, with more than 2.5 million downloads) and Instinct (which works over text message) browse merchant sites and complete purchases with the shopper's approval (Profound, PYMNTS). An agent that cannot select a variant or reach checkout moves on to another store.

• Run the test: Ask an agent (Muse or ChatGPT agent mode) to buy one specific variant, such as "the E-20 in stainless steel, delivered by December 20", and note the step where it stalls.
• Fix common blockers: Give variant swatches text labels (aria-label="Color: Black"), use standard buttons for add to cart, delay email sign-up pop-ups, and offer guest checkout. Check that fraud rules accept single-use virtual cards, which Muse uses at stores that don't accept Stripe's Link.
• Check access: Confirm that robots.txt and your CDN's bot protection (Cloudflare, Akamai) are not blocking AI crawlers and agent browsers, and review bot logs for user agents such as OAI-SearchBot, ChatGPT-User and Meta's crawlers.

An AI agent shops on a merchant site the way a first-time customer does, and it stops at the first step it cannot complete.

7. Run a weekly benchmark of unbranded holiday prompts through December

A fixed set of shopper prompts, run every week on major AI engines from now through December, will show whether your products appear in AI answers, whether their details are correct, and which of steps 1 to 6 to fix next (which GEO metrics to track). Leave your brand name out of the questions, since shoppers looking for gift ideas rarely name a brand.

• Build the prompt set: Write 20 to 30 prompts without your brand name across four types: recipient ("gift for a dad who grills, under $100"), deal ("best Black Friday air fryer deals"), deadline ("... that arrives before Christmas") and comparison ("E-10 vs E-20 for a latte drinker").
• Log the same fields each week: in ChatGPT, Gemini, Google AI Mode and Perplexity, record whether your product appears in the product cards, whether your store is the first offer, whether the price, deal and delivery date are correct, and which URLs are cited. In a ReFiBuy study, shoppers chose the first offer card 76% of the time (ReFiBuy).
• Map each gap to a fix: missing from product cards points to feeds (step 1), a wrong price to the offer (step 3), and competitors cited from publisher pages to coverage (step 5). For Google, Merchant Center's AI performance insights report your share of voice against competitors in AI Mode and AI Overviews.

Highest Priority Tasks

Adobe forecasts the largest rise in AI traffic on Thanksgiving Day, so feed, PDP and site fixes need to be live by mid-November.

Stellar assesses AI shopping readiness across the five pillars of AEO and prepares e-commerce sites for AI answer engines and personal shopping agents such as Muse and Instinct. To review your holiday catalog with us, email contact@stellar-ai.co.