Holiday AI Shopping Traffic Could Jump 130%: What Sellers Should Prepare

AI shopping traffic could jump 130% this holiday season. What should sellers prepare now?

Many cross-border sellers are asking the same thing: from Black Friday to Christmas this year, will more American shoppers simply ask an AI on their phone what to buy, and how can my store catch that traffic? The short answer: don’t rush to open new channels. First make sure AI can understand your hero products and is willing to recommend them, while you hold your ground on social and search.

According to Bloomberg, Adobe expects US holiday-season traffic to retail websites from AI assistants and shopping agents to grow 130% year over year. The same Adobe forecast puts total online sales at $275.1 billion, up only 6.7%. In other words, the pie is only a little bigger, but the doors leading to it are changing fast.

How is traffic from AI shopping assistants different from advertising?

Traditional advertising means buying a position: the higher you bid, the more likely you are to be seen. An AI shopping assistant works more like a sales associate who does the homework for the shopper. It reads the product information, reviews and discussions it can find online, then offers a few recommendations. You can’t buy that slot outright. You can only make your information clear and credible enough to get picked.

Adobe’s forecast reflects the shift: paid search traffic is expected to grow just 4% and social referrals 17%, while traffic from AI assistants is growing far faster than both. For sellers on a tight budget, this is actually good news, because AI recommendations depend on the quality of your information, not the size of your budget.

This work is usually called GEO (generative engine optimization): getting AI answer engines like ChatGPT, Perplexity and Google AI Overviews to mention and cite your brand and products in their answers.

Are ChatGPT ads worth it this year?

For most small and midsize sellers, it’s fine to wait this year. According to Modern Retail, ChatGPT’s ad business is growing: Sensor Tower counted more than 1,400 unique advertisers running ChatGPT ads in August, up 8% from the month before. But most of them are big companies, and Terence Einhorn of the measurement firm Measured said ChatGPT’s equivalent cost per click is roughly six times Google’s.

The report also notes that measurement firms question how much of the sales from these ads is truly incremental: many people who ask ChatGPT shopping questions were already planning to buy. Another source, Superfiliate co-founder Anders Bill, believes most brands haven’t yet moved meaningful budget away from Meta or search ads.

The takeaway is simple: if you don’t have a dedicated “learning budget,” you don’t need to rush into testing ChatGPT ads this year. You can compete for the same entry point by earning organic recommendations through GEO, and that doesn’t charge you per click.

Why do more channels make it harder to know what to do first?

Psychology has a classic finding called choice overload: when there are too many options, people find it harder to decide and are more likely not to decide at all. Sellers heading into peak season are in exactly this state. AI shopping assistants, ChatGPT ads, livestream selling, new platform subsidies: each one looks impossible to miss.

The usual result is touching everything and finishing nothing, then discovering on Black Friday that the hero product page still has typos and the return policy is vague. Too many choices lead to paralysis. The fix isn’t gathering more information; it’s deliberately cutting options. Keep only the three things that most affect your peak-season results, and get them right.

How much effort does each traffic source deserve this year?

The table is there to help you subtract: go all in on the first two rows, protect the third, and set the fourth aside for this year.

Five things to do before peak season

1. Pick 3 to 5 hero products. Peak-season traffic tends to concentrate on a handful of products, so your effort should too.
2. Write product pages AI can quote directly. Make the title say clearly what the product is and who it’s for. List specs, dimensions, materials, price, shipping time and return policy in short sentences or a table, not just inside images.
3. Add an FAQ that actually answers questions. Take the questions customer service hears most, like “What size space does it fit?”, “Is it machine washable?” and “How long does delivery take?”, and write them up as Q&A on the product page or your blog.
4. Walk through checkout on your phone. From opening the product page to completing payment, check load speed, price display and checkout steps. Whether traffic comes from AI assistants or social, it all lands on a page that has to work smoothly.
5. Keep English social posting going through Christmas, with no gaps. Plan your content calendar in advance and rotate gift guides, use cases and real buyer feedback, so there’s fresh, credible information about your brand outside the marketplaces too.

Black Friday falls on November 27 this year, about six weeks from now. Aim to finish steps 1 to 3 in the first two weeks and step 4 in week three, and start step 5 this week and keep it going through Christmas.

How should you rewrite product pages so AI recommends you more easily?

When an AI assistant answers a question like “storage boxes for a small apartment,” it needs facts it can compare directly. Sellers often do the opposite: they pile on adjectives and skip the specifics. Here’s a hypothetical before-and-after:

• Before: Premium Quality Storage Box, Perfect Gift, Best Choice for Your Home. All adjectives, and no facts AI can cite.
• After: Foldable fabric storage box, 13 x 13 x 11 in, fits standard cube shelves, removable washable cover, ships in 3 to 5 days, free 30-day returns. Size, fit, cleaning, shipping time and return policy are clear at a glance.

After rewriting, check one thing people often overlook: are price, specs and return policy consistent across your Amazon listing, your independent site and your social bios? If the same product is described differently in different places, AI can’t tell which version to trust and is less willing to recommend you.

Back to the five steps: steps 2 and 3 decide whether AI can understand you, and step 5 decides whether fresh content about you exists beyond the marketplaces. If no one on your team can keep writing natural English, consider ToucHigh’s Social Media Manager Autopilot. It produces US-English posts and articles in your brand voice and gives each one a predicted engagement score before it’s scheduled; you can approve posts one by one or let the system publish automatically. Product page and AI search audits and fixes are handled by GEO Autopilot. Both are described on the ToucHigh products page.