Black Friday: How ChatGPT and AI Engines Pick the Deals They Recommend (2026)

AIs do not relay your promotions: they verify them, then they choose

Black Friday 2026 has a new intermediary between your promotions and your customers. According to a KPMG study run by OnePoll among 3,000 UK consumers, 15% of shoppers plan to use an AI chatbot such as ChatGPT or Gemini to hunt for their Black Friday deals, and the share rises to 30% among 25-34 year olds (KPMG study, October 2025). But these assistants do not pass your offers along: they compose their own selection, crossing your pages with price comparison sites, buying guides and your price history. This article explains how that selection gets made, why some merchants monopolize the answers, and how to observe your category before November 27.

What shoppers really ask the AIs during Black Friday

The surprise is not that shoppers ask for deals: it is that they mostly ask for verification. 5WPR's analysis of 500 Black Friday prompts via Profound shows it: the top three uses are verifying a website's credibility (15.2% of queries), verifying trust in a brand (15.1%) and validating a discount (13.9%), meaning 44% of the volume is spent vetting unknown brands and recalculating strike-through prices (5WPR analysis, September 2026). "Is this a real discount?", "is this site legit?": questions shoppers used to ask a forum, or nobody, and now ask a machine that answers with your data and your competitors'.

During Black Friday, the AI is not a megaphone for your promotions: it is a verifier your customers send out as a scout. And that verifier has strong opinions: in 5WPR's recommendation tests, a pair of noise-canceling headphones was never recommended at full price or at 10% off, but was in 53% of trials at 40% off. The depth and credibility of the discount literally change the answer. The same severity applies to brand youth: out of 39 references to a brand less than a year old, 36 came with a warning or a caveat.

Where the deals AIs cite come from

Tests published by the trade press give a concrete view of the mechanism. In Tom's Guide's deal-hunting comparison, ChatGPT composes its answers from retailers and marketplaces (GameStop, eBay, GOAT, official sites), with direct links and price comparisons, while Perplexity leans more on price comparison tools (PriceCharting) and expert buying guides, suggesting checks across the big retailers (Tom's Guide test, January 2026).

Three families of sources stand out, and it is what we observe on our scans of deal queries too: your own pages, provided they are published ahead of time and machine-readable (price, discount and conditions in plain text); price comparison sites and aggregators, which give the engines the history needed to judge whether a discount is real; and the press's deal guides and selections, most of which freeze before the day itself. A merchant present on all three planes becomes hard to leave out of an answer; a merchant absent from all three simply does not exist in the conversation. And tempo matters: on "best [product] deal" queries, engines rely on what is online before the rush, not on pages that go live on Friday morning.

Getting chosen: the 5 conditions of a citable deal

1. A deal page published early and machine-readable. Your Black Friday page should exist now, with products, discount levels and dates in plain text in the initial HTML; the readability moves are in our product pages guide.
2. A real, verifiable discount. Engines cross-check your strike-through price against comparison sites and history; a fake discount detected means at best an absence, at worst an answer describing you as dubious. 5WPR's numbers confirm it: recommendation triggers on genuine discount depth.
3. Established brand credibility. Reviews in volume and freshness, third-party mentions, consistent information everywhere: that is what avoids the near-automatic warning reserved for brands without a track record. If your brand is young, this workstream comes before the promotions.
4. A presence where prices get compared. Comparison sites, deal aggregators, an up-to-date product feed: these are the sources where engines verify your discounts; being absent means letting the verification happen without you.
5. Conditions citable as is. Offer duration, order deadlines, extended return policy: simple sentences the AIs repeat word for word in their answers, like the rest of the year-end setup.

Observing your category before November 27

The most useful thing to do a month out costs nothing: the starting picture. List 10 to 15 realistic deal queries for your category ("best [product] deal", "[product] black friday", "is [your brand] black friday legit?") and test them across several engines, several times to smooth out variability. Note three things: who gets cited (and if it is always the same competitor, which sources they sit in that you ignore); from what (your pages, a comparison site, a press guide); and what is said about your prices and your trustworthiness. Run the same measurement again during the peak: the difference between the two pictures is the most honest measure of what AI Black Friday brought you or cost you, far more reliable than a single test. The volumes, for their part, are guaranteed: Adobe expects AI traffic peaks of +95% on Black Friday and +88% on Cyber Monday, after a 693% surge over the 2025 holidays.

A word for B2B and SaaS: the software Black Friday follows exactly the same rules, with "[category] black friday deal" queries asked by highly qualified buyers, less competition in the answers, and software directories playing the role of the comparison sites. An early, verifiable SaaS deal page is even more visible there than in B2C, precisely because few do it.

Frequently asked questions

Do people really use ChatGPT to find Black Friday deals? Yes: according to the KPMG/OnePoll study of October 2025, 15% of shoppers plan to use an AI chatbot to hunt for deals, and 30% among 25-34 year olds. And the use goes beyond hunting: 44% of the Black Friday prompts analyzed by 5WPR serve to verify brands, sites and discounts.

How does ChatGPT pick the deals it recommends? By crossing three families of sources: merchant pages published ahead of time (readable prices and conditions), comparison sites and price histories (to judge whether the discount is real), and the press's deal guides. The selection varies by engine and phrasing, hence the value of testing your own queries.

Can a fake discount hurt me in AI answers? Yes. Engines cross-check the strike-through price against comparison sites and history; 5WPR's analyses show recommendation only triggers on substantial, credible discounts, and that distrust signals (unknown brand, inconsistent price) produce explicit warnings in the answers.

When should I publish my Black Friday page to be picked up by AIs? Now. Engines compose their answers from pages already online and already crawled; a page published on the day arrives after the battle. Publish early, with dates and conditions in text, and update prices when the time comes.

How do I know if my competitors are being recommended instead of me? By testing your strategic deal queries across several engines, repeatedly, and recording who gets cited and from which sources. If the same competitor keeps coming back everywhere, the answer is in their sources: comparison sites, guides and reviews where they are present and you are not.

Does AI Black Friday concern B2B too? Yes: SaaS and service deals follow the same mechanism, with less competition in the answers and software directories in the role of the comparison sites. An early, verifiable B2B Black Friday offer gains disproportionate visibility there.

Conclusion

Black Friday has always been a battle for attention; it is becoming a battle of verification. The AIs advising your customers repeat neither your banners nor your superlatives: they read your pages, cross-check your prices and weigh your trustworthiness, then they name two or three merchants. You enter that shortlist with a page published early, a real discount, a corroborated brand and citable conditions; you steer it by measuring, before and during the peak, who gets cited on your queries. Things will keep moving between now and November 27, and we will update this article if the engines change the game. To take your starting picture today, the free Sapian Metrics audit tests your deal queries across 8 engines, in minutes, against your real competitors.