For thirty years, the shape of commerce online was basically fixed: a search bar, a list of results, a click, a cart, a checkout page. You optimised for it, budgeted around it, and built entire departments to win a spot near the top of it.
That shape is now coming apart. Not overnight, and not evenly — but the direction is no longer in doubt. Customers are starting to ask for what they want instead of searching for it, and in a growing number of cases, they’re buying without ever landing on your website at all. If you sell anything — a product, a service, a professional offering — this changes two things you’ve spent your career optimising: how you get found, and how the sale actually closes.
This issue looks at both halves of that shift, with the receipts to back it up, and what it practically means for you as a business leader — regardless of your size or sector.
The front door moved.
Search hasn’t disappeared. But it’s no longer the only, or even the first, door customers walk through. Industry data suggests traditional search volume is on track to fall by roughly a quarter, with a meaningful share of that traffic migrating to AI-driven discovery — chat assistants, AI-powered answer engines, and conversational shopping tools.
Consumer research firm Yext puts the number of shoppers now using AI search tools daily at 43%, and 62% say they trust AI for brand-related decisions.
It’s also becoming category-specific rather than uniform. Research from Fractl found ChatGPT has become the second-most-used destination for health questions, and ranks second or third for product research, travel planning, and general how-to content.
There isn’t one AI platform to “optimise for” the way there was one Google to optimise for — different categories of intent are settling on different tools.
Here’s the part worth sitting with, though: none of this means customers are simply handing over trust to the first answer an AI gives them. Yext’s research shows the average shopper now checks roughly 2.4 different platforms before deciding — and that number holds steady from Gen Z to Baby Boomers. People are using AI to widen the funnel, not to skip the thinking.
Checkout is being rebuilt — and the standard isn’t settled yet.
The more consequential change is happening a step further down the funnel: at the point of payment.
In late 2025, OpenAI began letting shoppers buy directly inside ChatGPT through a feature called Instant Checkout, launching first with Etsy and Shopify, then Walmart. At the same time, Google introduced its own Universal Commerce Protocol at the NRF 2026 retail conference, and Microsoft unveiled Copilot Checkout for its own AI assistant. Mastercard and, later, Visa announced they’d plug their payment networks directly into these AI systems, effectively building the plumbing for an “agent” to pay on a customer’s behalf.
It hasn’t gone entirely to plan. By March 2026, reporting on OpenAI’s own retail partnerships noted the company was pulling back from fully in-chat checkout and shifting toward dedicated in-app experiences that route the customer back to the retailer’s own website to complete the purchase.
A Gartner analyst’s assessment at the time was blunt: enabling AI-driven transactions had proven harder than anyone anticipated — for OpenAI and for the retailers alike.
The retailer that seems to have read this best is Walmart. Rather than ceding the entire transaction to a third-party chat interface, it built its own AI shopping assistant, Sparky, and embedded it into ChatGPT — keeping its brand and customer relationship intact even as the discovery happens somewhere else.
Walmart has reported that ChatGPT is now driving roughly twice as many new customers as traditional search referrals. Sephora took a similar approach, launching a guided AI shopping experience inside ChatGPT rather than opening its full catalogue to an outside checkout flow.
Then there’s Amazon, which has taken the opposite stance entirely: it has actively blocked third-party AI agents, including ChatGPT, from shopping on its own site, while building out its own assistant, Rufus.
That held even after Amazon announced a strategic partnership with OpenAI in February 2026 involving up to $50 billion in investment — a deal centred on cloud infrastructure and model access, not shopping integration. As of this writing, Amazon’s shopping experience remains its own walled garden.
The takeaway for the rest of us isn’t “get on every platform immediately.” It’s that the businesses navigating this well are the ones deciding deliberately how much of the customer relationship they’re willing to hand to someone else’s AI — and building their own presence inside these tools rather than simply hoping to be found by them.
Adoption is outrunning trust.
Here’s the tension every executive should hold onto before reading too much hype into any of this: usage of AI in shopping is climbing fast, but trust in it is not keeping pace — and in some measures, it’s moving backwards.
Fractl’s 2026 consumer survey found the share of people who say heavy AI use by a brand would reduce their trust in that brand doubled year over year, from 20% to 40%. In the same study, the share of consumers who rated AI as more helpful than traditional search fell from 82% to 54% — a 28-point drop in twelve months, as the novelty wears off and people run into generic or low-confidence answers.
Separate research from Klaviyo found only 13% of consumers say they “completely” trust AI, with the rest split between cautious and neutral.
There’s a useful, related data point from a Canva-commissioned marketing study: 68% of consumers said they don’t mind AI in advertising if it makes the ad more genuinely useful — but 78% said they’d still rather see content made by a person, even if the AI version might technically perform better. Efficiency and preference are pulling in different directions.
None of this means AI adoption stalls. It means the businesses that win the next few years won’t be the ones that moved fastest — they’ll be the ones customers still trust once the novelty has worn off everyone else.
What this means for your business.
You don’t need an AI checkout button on your website next quarter. Most mid-sized and smaller businesses shouldn’t chase that yet — the standards themselves (Google’s UCP, OpenAI’s protocol, Microsoft’s Copilot Checkout) are still competing for dominance, and building for the wrong one is wasted effort. But there are three things worth doing now, regardless of your size:
1. Make yourself legible to AI, not just to Google. AI agents and answer engines can only recommend what they can parse cleanly — accurate, structured, consistently updated product and service information, on your own site and everywhere else you’re listed. This is quietly becoming as important as SEO ever was. Audit where your business’s information is inconsistent or stale; that inconsistency is now costing you visibility in a channel you can’t see.
2. Don’t assume disintermediation is inevitable — decide your posture on purpose. Walmart’s and Sephora’s approach — build your own assistant, place it inside the platforms your customers already use, keep the relationship — is a more defensible strategy for most businesses than either full resistance (Amazon’s approach, viable mainly if you have Amazon’s scale) or full surrender (handing your entire funnel to someone else’s checkout).
3. Earn trust deliberately; don’t assume adoption implies it. If you’re using AI in customer-facing marketing, service, or recommendations, be transparent about it and keep a human in the loop where the decision matters.
The data is unambiguous: your customers are watching for exactly the moment you let AI substitute for genuine care, and they are less forgiving of that than they were a year ago.
The bigger picture is this: the funnel hasn’t disappeared; it’s fragmented — and the businesses that treat that fragmentation as infrastructure to get right, rather than a marketing trend to chase, will be the ones still standing once the dust settles on which “checkout protocol” wins.
