Black Friday in the Age of AI: How Shopping Journeys Have Changed

How shoppers are using ChatGPT, Google AI Mode and Perplexity to research deals

Typing three keywords into Google and scrolling through ten blue links is on the way out for a big chunk of shoppers. What's replaced it: asking a proper question and getting a shortlist handed back. Ask “what's the best noise-cancelling headphones deal right now” in ChatGPT, Google's AI Mode, or Perplexity, and you get a completely different kind of answer to what a traditional search box gives you (and often a different answer from each of the models!)

The numbers back this up. 49% of consumers tried AI-assisted shopping at some point in 2025, and 64% say they'll use AI chatbots for shopping in 2026, with ChatGPT the clear favourite at 61% preference over the other tools. It skews hard by age too: 54% of Millennials likely to use AI for holiday shopping, against 19% of Boomers.

One caveat I'd give any client assuming AI Overviews have taken over shopping search: they haven't, at least for transactional queries. Coverage on e-commerce queries actually dropped to around 4% by February 2026, down from 29% a year before, because Google's protecting its Shopping ad revenue. Where AI Overviews dominate, however, is the informational, comparison-led queries further up the funnel (best X for Y, is X worth it), which is exactly where the research and consideration happens before someone's ready to buy.

What's actually changed is who sets the default option a shopper sees first. It used to be whatever a retailer's homepage or a Google ad put in front of them. Now it's whatever the AI's shortlist surfaces. If your brand isn't on that shortlist, you've effectively been opted out before the shopper has consciously chosen anything.

Owning the Google Shopping surfaces: Merchant Center and Ads promotions

That last point is worth sitting with. Google is deliberately keeping AI Overviews away from transactional queries to protect Shopping ad revenue, which means Merchant Center and Google Ads promotions are still doing most of the heavy lifting when someone's actually ready to buy. 

Merchant Promotions and Google Ads promotion assets are two separate systems, and mixing them up costs people time every single year. Merchant Promotions run through the product feed itself, using the “sale_price” and “sale_price_effective_date” attributes to get a strikethrough price and badge showing on Shopping listings - these trigger the mental reaction to loss aversion - people don’t want to miss out on a deal! Promotion assets, by contrast, sit on Search and Performance Max campaigns: you set an occasion, a discount type and some short item text in Assets, and it shows as a bold callout next to your ad without you touching the ad copy itself.

Google's own guidance is blunt about timing, and it's worth repeating because it catches people out every year: don't submit promotions at the last minute. Review can take 48 hours or more once everyone else is submitting Black Friday offers at the same time, so a promotion queued up on the Wednesday before doesn't always go live in time. Get sale pricing and promotion assets approved and scheduled well in advance, build dedicated seasonal Shopping campaigns around your best sellers for the extra visibility, layer remarketing and similar-audience lists across them, and check your inventory filters aren't quietly excluding products from being shown at all.

A live, clickable promotion badge next to a price still does real work; it's one of the oldest tricks in the book for making a deal feel urgent and worth grabbing. It only works if it's actually there when the shopper looks, though. A disapproved promotion or a stuck review does the same damage as never running the offer at all.

The rise of delegated shopping decisions

Researching with AI is one thing. Handing it the actual decision is another, and shoppers are edging that way too. 34% say they're comfortable with an AI agent automatically completing a purchase based on their stated preferences and budget, and 39% say they're at least somewhat likely to buy something an AI recommends. 13% would let AI complete a purchase with no review at all.

Mastercard's forecasting puts a number on where this ends up: by 2030, more than 1 in 10 online shoppers (over 300 million people globally) will routinely use AI agents to shop and pay, starting with low-risk, repeat-purchase categories like groceries, medicines and subscriptions rather than big-ticket or gifted items. Worth watching too: 27% of teens say they'd use fully autonomous AI shopping tools, against 16% of parents, and 31% of teens say they trust AI product advice over their friends' opinions.

Something which is very important to know: OpenAI's early push into full “Instant Checkout” inside ChatGPT has already been walked back in favour of a discovery-first model. 

Social signals: shaping the decision before the journey even starts

Here's the bit that gets missed when the whole conversation is about chatbots: a huge amount of purchase influence happens before a shopper's even consciously “researching” anything. 85% of Gen Z shoppers say social media influences their purchase decisions, with TikTok and Instagram named the biggest influence by 45% of that group, ahead of YouTube at 38%. That influence gets seeded months before Black Friday, through UGC, creator content and organic reach, not in the days around the sale itself.

Where this gets genuinely interesting for paid and organic social is the overlap with what AI tools cite. Reddit accounts for roughly 21% of all Google AI Overview citations and YouTube for 23.3%, both now significant sources that large language models pull from when they build an answer. That means a Reddit thread arguing over the best air fryer in October, or a YouTube unboxing that picks up traction in early autumn, can end up shaping what an AI recommends when someone asks “what's the best Black Friday deal on an air fryer” in November, even though that shopper never saw the original post - social proof is evolving in our modern AI-first world.

Repeated, early exposure just makes a brand easier to recall, and easier for an AI system to find and cite, by the time high-intent search happens. If you're waiting until the last week of November to push social activity, you're missing the window that actually matters, the one most brands still treat as off-season.

Organic reach only stretches so far, and that's where paid social earns its keep: it lets you choose when and who sees that seeding activity, rather than hoping the algorithm does it for you. The three platforms worth putting budget behind each do a slightly different job.

Meta: Advantage+ and creative diversity

Meta is leaning harder into automation with Advantage+ Shopping campaigns, handling budget and targeting algorithmically rather than at ad-set level. That shifts the main lever you've got left onto creative diversity: feed the system more formats and angles rather than trying to out-bid it. The newer buy-now button inside ads works well for impulse purchases under about £50, but anything pricier still needs a proper landing page rather than a one-tap checkout.

TikTok: phased testing and Spark Ads

TikTok recommends a phased run-up rather than a standing start: testing creative through September and October, then scaling into early November to catch both early planners and last-minute shoppers. Spark Ads (boosting an existing organic video or creator post while keeping its comments and shares intact) hold onto social proof in a way a fresh ad doesn't, and pairing them with Catalog Ads for in-feed browsing is where TikTok's own case studies show the biggest efficiency gains.

Pinterest: catching intent before it has a brand attached