Demand Generation

AI Starts the Holiday Shop, but Trust Still Stops at Checkout

Bain expects US holiday sales to top $1 trillion for the first time, and 24% of online shoppers plan to start on an AI platform. Yet only 23% trust AI to handle a payment, which leaves the last step of the journey firmly with the brand.

September 23, 2026 · Demand Generation
A laptop, two payment cards, a coffee cup and a parcel of folded clothes on a pink desk

Key Takeaways

  • Bain forecasts November and December US retail sales above $1 trillion for the first time, with more than half of the 4.5% growth coming from inflation.
  • 24% of online shoppers plan to start their holiday shopping on an AI platform, up from 17% in 2025, and a further 13% plan to use retailer AI agents.
  • Only 23% of US consumers trust generative AI to handle a payment on their behalf, rising to 61% when a trusted payment brand is involved.
  • Of the 40% of shoppers open to using AI this season, 42% say they will not let it buy a gift for them.

The 2026 holiday season is about to break a record that says less than it seems to. Retail sales are heading past a trillion dollars, a growing share of shoppers will open the season by asking an AI assistant what to buy, and almost none of them want that assistant anywhere near their wallet. For demand teams, the result is a journey that starts somewhere they do not control and ends somewhere only they can.

A Trillion-Dollar Season Built on Deliberate Shoppers

The Bain & Company holiday forecast, published on September 3, puts November and December US retail sales above $1 trillion for the first time, growing 4.5% in nominal terms against 3.5% in 2025. More than half of that growth comes from higher prices rather than more buying. In-store sales are expected to grow 2.5% and nonstore sales 9%, with nonstore channels delivering 60% of the season's growth even though stores still account for around 70% of sales.

Deloitte's forecast, released a week later on a wider November to January window, lands in the same place: total holiday sales of $1.70 trillion to $1.71 trillion, up 4.0% to 4.8%, and e-commerce of $316.1 billion to $318.9 billion, up 7.5% to 8.4%. Deloitte economist Akrur Barua credits online growth to shoppers using digital tools to research, compare and complete purchases, while vice chair Natalie Martini describes value-seeking behaviour across income levels, including switching among brands and retailers.

That is the tension underneath the headline. A record season made partly of inflation, bought by shoppers who compare harder and switch more readily, rewards the brands that are present at the moment of comparison. Bain's Aaron Cheris frames the retailer's task as striking the right balance on price and promotions, and with roughly 90% of shoppers planning to take part in a major sale event, the promotional calendar will be crowded early.

It is also a longer season. Basis's holiday trends survey of 2,006 US consumers found 68% shop early to avoid delays and 57% shop year-round for deals, stretching the buying window from September into January. The comparison phase is no longer a week in late November. It is already underway.

The AI Assistant Now Owns the First Question

Bain's survey of more than 1,100 consumers found 24% of online shoppers plan to begin their holiday shopping on an AI platform such as Claude, Google Gemini or ChatGPT, up from 17% last year. Another 13% plan to use a retailer's own AI agent. Retailer and brand websites are rising too, used by 60% of shoppers against 51% in 2025, which suggests AI discovery is adding a step to the journey rather than replacing the destination.

Basis puts the broader picture at 71% of consumers using at least one digital tool, whether search, social or AI, to find gift ideas last year. For a demand team, the practical consequence is that the shortlist is increasingly assembled before the shopper reaches an owned channel. A brand that is invisible to an assistant, or described badly by one, can lose the sale before its own campaign has a chance to run.

Then the Shopper Takes the Wheel Back

Delegation collapses at the point of payment. Speaking at an investor conference on September 8, Visa chief executive Ryan McInerney said three-quarters of consumers surveyed did not trust agentic platforms to make payments autonomously, and summed up the state of the market in one line reported by PYMNTS: “We are seeing adoption for shopping, but not yet for autonomous payments.”

The Visa research released the following day, a Harris Poll survey of 2,065 US adults fielded in May, gives the detail. 72% of consumers have used an AI assistant, but only 23% trust generative AI to handle a payment transaction on their behalf. That figure rises to 61% when a trusted payment brand is involved, and to 71% among frequent AI users. Visa is an interested party here and the brand-specific number should be read that way, but the direction of travel is the useful part: trust in the payment layer is doing work that trust in the AI layer cannot.

The same split appears in Basis's data. Of the 40% of shoppers open to using AI this season, 42% say they will not let it buy a gift on their behalf. PYMNTS reporting on consumer comfort finds 56% willing to let agents search and compare products, but only 35% willing to let them access saved payment credentials.

So the journey this season has a strange shape. The first question is increasingly answered by a machine, and the last decision is emphatically human. Everything that happens between those two points, on the product page, in the basket and at checkout, is where the brand still gets to make its case, and where a slow, unfamiliar or untrusted payment step can undo everything the discovery phase earned.

A Holiday Playbook for the Split Journey

The assistant will increasingly decide which brands make the list. Whether the shopper actually buys is still decided in the last few seconds, by a person, on a page the brand controls, and that is where holiday budgets can still earn their keep.

Share

More in Demand Generation

All Resources →