AI shopping boom puts checkout under pressure

By now, it is widely accepted that AI is changing how people shop. Search is faster, recommendations are more relevant, and consumers can move through options with less effort than before.

What matters more now is what happens next.

As AI gets better at helping people search, compare and decide, it also changes where conversion is won or lost. From a merchant's perspective, that is the more important shift. The front end may be getting smarter, but that does not automatically lead to better conversion. In fact, it often does the opposite: it makes weaknesses in the transaction layer much more visible.

That is why agentic commerce deserves attention. Its impact does not stop at discovery. It extends all the way to whether a transaction is actually completed. Once AI starts doing more of the work that used to sit with the consumer, whether filtering options, narrowing choices or shaping the final decision, payments, authentication and authorisation stop being quiet back-end functions. They start influencing conversion much earlier and much more directly.

There are already clear signals of this. Adobe's latest data shows that in the first quarter of 2026, AI-sourced traffic to U.S. retail sites rose 393% year over year, while AI-sourced traffic to travel sites rose 233%. At the same time, 39% of consumers said they had already used AI assistants for online shopping, and 85% of those users said the experience improved shopping for them. In travel planning, that figure rises to 86%.

That is worth paying attention to. AI is not just bringing in more traffic. It is bringing in traffic that is more informed, more intentional and closer to purchase.

For many merchants, that changes the conversion logic they have been working with for years. The traditional focus has been on visibility, recommendation and getting the user onto the site. But when AI starts compressing part of the search and decision process on the consumer's behalf, the transaction layer becomes more exposed. The shorter the path to purchase, the less room there is for checkout friction, broken authentication flows or inconsistent payment experiences.

This is what I think is often missed in discussions around agentic commerce. The bigger question is not simply how much efficiency AI can unlock for merchants. It is which parts of the journey, previously treated as secondary, that suddenly become central.

Checkout is the clearest example. Baymard still puts the average documented cart abandonment rate at 70.22%. Even in a traditional eCommerce journey, the transaction stage is already fragile. If AI now compresses the decision-making process even further, any friction that remains during payment becomes more visible and more costly.

Many merchants have historically treated payments as an execution layer. If funds can be collected, transactions can be processed, and local payment methods are supported, the job is considered done. But in a shorter, AI-shaped decision journey, that view starts to fall apart. Payments are no longer just operational plumbing. They become part of the experience itself.

If the customer journey begins with a fluid, conversational interface and ends with a cluttered checkout page, multiple redirects for authentication or a fragmented local payment experience, what the user feels is not efficiency. It is an interruption.

Travel is a particularly useful lens for understanding this shift. It is a high-complexity category, shaped by multiple variables, higher-consideration decisions and frequent cross-border payments. That makes it especially well-suited to AI-assisted decision-making. It also means transaction bottlenecks tend to surface faster.

AI can help narrow down destinations, budgets, dates and options much more quickly. But if the payment and authentication logic behind the journey cannot keep pace, the efficiency gained earlier in the flow is easily lost at the last mile.

This is one of the areas my team and I have been watching closely at Unlimit. In one OTA use case we have explored internally, the journey begins with conversational trip discovery, moves into option selection, then into passkey-based authorisation and instant booking confirmation. The point is not the specific solution itself. What matters is the broader direction it reflects: as AI becomes more deeply involved in shaping user decisions, payments and authentication need to be designed into the journey much earlier, not added at the end once discovery and comparison are already done.

That does not mean fully agentic transactions are about to scale overnight. McKinsey's 2026 research points to the other side of the story: close to two-thirds of companies have experimented with AI agents, but fewer than 10% have scaled them in a way that delivers meaningful value. Data foundations, governance and operating models remain real constraints.

That is exactly why the most practical priority for merchants today is not chasing a fashionable concept. It is strengthening the core capabilities that will actually determine whether conversion holds up as buying journeys evolve.

The next competitive edge may not come from having AI capabilities alone. It may come from having the transaction infrastructure that can support AI-shaped decision journeys without adding friction, losing trust or breaking the flow at the moment of purchase.

AI will keep getting better at shortening the distance between intent and decision. The businesses that will benefit most are the ones that can turn that shorter path into a completed transaction.