A parent types a question into an AI assistant late one evening: "Best gift for a ten-year-old who loves science, under $50, that will arrive before Christmas?" A few seconds later, the answer lists four products, a sentence on why each fits, and where to buy them. A small brand makes a science kit that is better reviewed and cheaper than two of the four. It isn't mentioned. The parent never sees it, never searches for it, and buys one of the others.
That exchange is happening at growing scale this season. Adobe's 2026 holiday forecast, published on 28 September, expects AI traffic to US retail sites, measured as shoppers clicking a link through to those sites, to rise 130% year on year. With Thanksgiving three weeks away, the uncomfortable truth is that brands can't rebuild their authority in time. But they can still change a lot. Before peak, work on what AI retrieves, not what it remembers. The sources assistants look up live, from product data to fresh reviews to gift guides, can change within days or weeks. What models learned in training can't.
This post is a short-term action plan for the weeks before peak. How AI is changing gift discovery more broadly is covered in our upcoming holiday search guide, and the longer-term shift in agentic commerce in 2026.
What the numbers say, and what they don't
Adobe expects US online holiday spending to reach $275.1 billion across November and December, up 6.7%, with AI traffic to retail sites growing 130% over the season and the biggest uptick expected on Thanksgiving. In Adobe's survey of 5,000 consumers, fielded in July 2026, 77% of people who had used AI for online shopping said it made them more confident in their purchase, and 69% said they were less likely to return an item bought with AI's help. AI tools are also adding shopping features: ChatGPT's shopping research option can build personalised buyer's guides, and Google's Universal Cart, rolling out in the US across Google Search and the Gemini app, can send notifications about price drops and items coming back in stock.
Two cautions keep this in proportion. Adobe counts shoppers clicking a link to a US retail site, so the figure isn't a measure of total AI shopping activity or of all AI-assisted purchases. And Adobe hasn't disclosed AI's current share of retail visits, so 130% growth may still be a small slice of total traffic for many brands. The reason to act anyway is our reading, not Adobe's data: someone who clicks through from an AI answer has often already narrowed the options, so these shoppers may arrive later in their decision, with a shortlist that you're either on or not.
Flocksy
Unlimited flat-rate creative subscription for graphic design, video, copywriting and web work.
Best for: Businesses wanting unlimited flat-rate design & creative services
Where AI gift answers come from
How assistants answer shopping questions varies by assistant and product experience: many combine live web information, merchant product data and other public sources with what the model already knows. OpenAI's help page on shopping in ChatGPT, for example, says it considers structured product data from first- and third-party providers, other third-party content and the model's own responses. That retrieved layer is what makes short-term work possible. The timings below are indicative planning estimates, not measured timelines; actual timing depends on the platform, crawling, feed refreshes, indexing and when sources are updated.
| Source | Check before peak | Indicative timing |
|---|---|---|
| Product feeds and data | Prices, stock, identifiers, shipping | Days |
| Your product and gift pages | Clear who-it's-for, budget, delivery dates | Days to weeks |
| Reviews and ratings | Recent volume and specific detail | Weeks |
| Gift guides and best-of lists | Inclusion in lists still being updated | Weeks |
| What the model remembers | General brand reputation | Months |
The last row is why a pre-peak sprint shouldn't chase brand-level fame. Shaping how models understand your company is valuable work, covered in keeping AI accurate about your brand, but it pays off next year, not next week.
The real deadline is earlier than you think
Changes don't appear in AI answers the moment you publish them. Crawlers need to revisit pages, feeds need to refresh, and new reviews and lists need to be picked up. That lag moves your practical deadline forward.
From 4 November: Thanksgiving is 22 days away, Black Friday 23, Cyber Monday 26.
If you assume changes take roughly 1–2 weeks to be picked up and reflected in answers:
→ With a 2-week lag, fixes need to be live by Thursday 12 November
→ With a 1-week lag, by Thursday 19 November
That leaves about one to two weeks of working time. Adobe expects AI-referred traffic growth to peak on Thanksgiving itself, so work that lands after mid-November risks missing the days that matter most.
Lags vary by assistant, site and source, so treat these as planning assumptions, not a promise that changes will appear in AI answers within that window.
The rule for the next two weeks Before peak, work on what AI retrieves, not what it remembers. Product data, gift pages, fresh reviews and gift guides can change within days or weeks. Brand reputation in a model's memory can't.
A two-week sprint to improve your chances of being cited
None of this guarantees a mention, but each step makes it more likely that assistants can find, understand and recommend your products for the right questions.
1. Ask the questions your buyers ask. Holiday prompts usually combine a recipient, an occasion, a budget and a constraint: "gift for a dad who likes hiking, under $75, that ships by the 20th". Write 20 to 30 of these for your categories, using the approach in question research for AI search, and run them in ChatGPT, Google's AI features, Gemini and Perplexity. Note which products and sources appear, whether you're included, and whether details about you are correct. Repeat important prompts, since answers vary between runs, or use the tools covered in AI visibility platforms.
2. Fix your product data first. It's the fastest lever. Make sure feeds and product pages carry accurate prices, stock levels, product identifiers, variants, shipping costs, holiday delivery cut-off dates and returns terms. An assistant matching "arrives before Christmas" can only include you if that information exists and is right. The practical details are in shipping and delivery promises and product page optimisation.
3. Write for the gift question, not just the product. Assistants answer "who is this for and why?" Add clear, quotable passages to product pages or a dedicated gift hub: who the product suits, what problem it solves, the age range or skill level, the price band, and delivery-by dates. Plain, specific sentences are easier for assistants to use in answers, as covered in writing pages AI tools quote directly.
4. Top up recent reviews. Many assistants draw on reviews to judge quality and fit, and recent, detailed reviews give them more to work with. Make sure October and early November buyers receive a well-timed, fair request, using the approach in asking for reviews after purchase.
5. Get into the lists AI cites. Look at which gift guides, publisher round-ups and forum threads appear as sources for your prompts. Many publishers update gift guides through November, so offer samples, clear product information and genuine reasons for inclusion. This is the fast version of the longer strategy in earning third-party mentions.
6. Make sure AI can reach you. Access requirements differ across search engines and AI services, and no single setting controls AI visibility, so check each relevant service's documented crawler and robots.txt guidance to make sure you aren't accidentally blocking it. Also check that key pages load quickly and that important information isn't hidden behind scripts or pop-ups that some crawlers may not process.
During peak: monitor and correct
Once the season starts, shift from building to watching. Re-run your key prompts every few days, and correct any errors quickly: wrong prices, out-of-stock products being recommended, or outdated delivery promises. Make sure the pages AI-referred shoppers land on confirm what the assistant told them, the principle behind landing pages for visitors sent by AI. Track AI referral traffic in your analytics by landing page, and note which prompts and products performed, because next year's work starts with this season's evidence. For the broader seasonal plan, see the 90-day Black Friday timeline. Brands that want help turning AI visibility into a repeatable programme often work with a specialist search engine optimization team.
The short version
Adobe expects AI traffic to US retail sites, measured as clicks through from AI tools, to rise 130% this holiday season, with the biggest uptick on Thanksgiving, and its survey found most AI shoppers felt more confident and less likely to return items, though AI's share of total traffic is undisclosed and may still be small. Many AI shopping answers draw on sources retrieved live, though methods vary by assistant, and that is what makes short-term work possible. Before peak, work on what AI retrieves, not what it remembers: product feeds, product and gift pages, reviews and gift guides can change within days or weeks, while brand reputation in a model's memory takes months. Because changes take time to register, aim to have key fixes live by mid-November, 12 November with a two-week lag. In a two-week sprint, run your buyers' gift prompts across assistants, fix product data and delivery dates first, write clear who-it's-for passages, top up recent reviews, pursue inclusion in gift guides assistants cite, and check each service's crawler guidance so you aren't blocking access. During peak, re-run prompts, correct errors fast and track AI referrals.
Is your brand showing up in AI gift answers?
We help brands improve how their products show up in AI shopping answers before and during peak.
Explore SEO →Frequently asked questions
How much will AI influence holiday shopping in 2026?
Noticeably more than last year, though from a base that is still small for many retailers. Adobe's 2026 holiday forecast, published on 28 September, expects AI traffic to US retail websites, measured as shoppers clicking a link through to those sites, to rise 130% year on year across November and December, with the biggest uptick on Thanksgiving, within a season Adobe expects to reach $275.1 billion in US online spending. In Adobe's survey of 5,000 consumers, fielded in July 2026, 77% of people who had used AI for online shopping said it made them more confident in their purchase, and 69% said they were less likely to return an item bought with AI's help. Because Adobe counts clicks through to retail sites, the figure is not a measure of total AI shopping activity or all AI-assisted purchases, and Adobe has not disclosed AI's current share of retail visits, so the growth is fast but may represent a small share of total traffic. Our reading is that shoppers who click through from AI answers are often further along in their decision, so the questions they ask can shape which products get considered.
Where do AI assistants get their holiday product recommendations?
From a mix of sources that varies by assistant, product experience and question. Many shopping experiences draw on live web information and merchant product data, and may use retailer and brand product pages, product feeds, reviews and ratings, third-party gift guides and best-of lists, publisher reviews, and discussions on sites such as Reddit and YouTube, often combined with what the model already knows. No assistant necessarily uses all of these, and exact methods are not always disclosed. Structured information such as price, availability, shipping and returns details helps assistants match products to constraints like budget and delivery dates. Separately, what a model learned during training shapes its general sense of brands and categories, but that knowledge is older and changes slowly. For brands trying to influence holiday answers in a matter of weeks, the most practical levers are the sources assistants may retrieve live: accurate product data, clear product and gift pages, fresh reviews, and presence in the lists and guides that assistants cite.
Is it too late to optimise for AI shopping before Black Friday?
Not entirely, but the window is short and the work must be chosen carefully. Building broad brand authority or changing how AI models remember your brand takes months, so that is not realistic before peak season. What can still move within weeks is information that AI assistants may retrieve live: product feeds, product and gift pages, prices and stock, shipping and returns details, fresh reviews, and your presence in gift guides that are still being updated. Because assistants and search systems need time to re-crawl and refresh pages, a sensible planning assumption is to have changes live one to two weeks before the days that matter most, though actual timing varies and isn't guaranteed. For Thanksgiving on 26 November 2026, that points to finishing the most important fixes by mid-November. After that, focus shifts to monitoring: checking what assistants say about your products, correcting errors quickly and making sure landing pages serve AI-referred visitors well.
How can you check whether AI assistants recommend your products?
Ask them the questions your customers ask. Write a set of realistic gift and shopping prompts that combine a recipient, an occasion, a budget and a need, such as the best gift for a teenager who likes photography under a certain price, and run them in the assistants your customers use, such as ChatGPT, Google's AI features, Gemini and Perplexity. Record which products and sources appear, whether your products are mentioned, and whether the details given about them, such as price, features and availability, are correct. Because answers vary between runs, repeat important prompts several times rather than relying on a single check. AI visibility tools can automate this across many prompts and assistants. In your analytics, track referral traffic from AI platforms to see which pages AI-referred shoppers land on and how they behave. Use what you find to prioritise fixes to product data, pages and third-party presence.
