Retailers are understandably excited about the potential of Agentic Commerce.
AI-powered shopping assistants can help customers discover products, compare options, answer questions and complete purchases. Personalisation is becoming more sophisticated, search is becoming more intelligent and the potential to reduce friction and improve conversion is significant.
From a revenue perspective, I completely understand why this has become such a major focus.
Every retailer wants to make it easier for customers to find and buy the products they want. If AI can improve conversion rates, increase basket sizes and create better customer experiences, there is a clear commercial opportunity.
But I think we’re at risk of looking at the opportunity too narrowly.
Because while the industry is focused on using AI to help retailers win the order, we should also be asking how it can help them protect the value of that order once it has been placed.
After all, revenue doesn’t end at checkout.
The customer journey doesn't end when the sale is made
Retailers invest heavily in getting customers to the point of purchase. They invest in marketing and customer acquisition. They optimise websites and checkout experiences. Increasingly, they’re investing in AI to make the path to purchase faster and more seamless. But once an order is placed, the customer journey moves into eCommerce fulfilment and delivery.
But once an order is placed, the customer journey moves into fulfilment and delivery.
That’s where things get harder to control. Stock has to be available, the order has to be picked and dispatched, a carrier has to perform, and the parcel has to arrive when the retailer promised. Any of those can go wrong: a stock shortage, a fulfilment bottleneck, a carrier disruption, a missed delivery. Operationally, these are everyday challenges. Commercially, they’re a lot more than that.
A retailer can invest significantly in acquiring a customer and optimising the journey to checkout, only for a poor post-purchase experience to undermine everything that came before it.
That is why I believe one of the biggest blind spots in the Agentic Commerce conversation is what happens after the customer clicks buy.
Who is building AI for the moment of failure?
Much of the conversation around Agentic Commerce has focused on the front end of retail. How can AI help customers discover products? Make better recommendations? Answer questions? Complete transactions?
But the more interesting question for me is this: Who is building AI for the moment something goes wrong?
Traditional systems are good at spotting problems. They flag unavailable stock, surface delivery delays, show when a carrier is disrupted. Spotting a problem and fixing it are not the same job, and that gap is where agentic AI earns its place.
Imagine an order cannot be fulfilled from its original location. Rather than simply generating an alert and waiting for human intervention, an AI-powered system could potentially identify alternative stock, assess fulfilment options, evaluate carrier capacity, adjust delivery estimates and execute a new plan. That, in practice, is what good delivery management software should do: not just report the exception, but work out the best response to it.
This is starting to move from idea to plan. In our latest research, 58% of UK enterprise retailers say they’re likely to hand an automated AI system real control over delivery strategy and carrier operations. Ask them where it will earn its keep and they point at exactly these moments: automating order allocation to final-mile carriers (34%), predicting and preventing failed deliveries (33%), managing the technical integrations underneath (33%), and monitoring delivery problems and carrier performance (31%). That kind of dynamic reallocation only works when a retailer has real multi-carrier shipping options to move between, and it’s a clear sign retailers want AI making the operational call itself.
Operational problems are commercial problems
I think this is where revenue and operations need to become much more closely connected.
Businesses often treat commercial performance and operational performance as separate conversations. One team focuses on acquiring customers and driving growth. Another focuses on fulfilling the order.
But customers don’t see it that way.
They don’t separate the website from the warehouse or the purchase experience from the delivery experience. They remember the experience they had with the brand.
And when delivery goes wrong, the commercial consequences can be significant.
Scurri research has found that 84% of consumers would stop buying from a brand after a single poor delivery experience.
That should make us think differently about post-purchase.
Delivery isn’t simply the final operational step in completing an order. It’s part of the customer relationship.
And protecting that experience is also about protecting future revenue.
There’s a catch, and it’s the other half of the same research. While 58% of retailers are ready to hand AI the controls, only 37% of consumers are comfortable with AI making decisions about their deliveries. 88% want the option to reach a human, 64% suspect retailers are turning to AI mainly to cut costs, and just 33% trust AI to use their data responsibly.
So the commercial risk runs in two directions. Get delivery wrong and you lose the customer. Automate it in a way that feels opaque or hands-off, and you can lose them just as quickly. The encouraging part: 60% would happily use an AI agent that fixes a delivery problem on its own, as long as they keep a say in how it works. The technology has permission. It just comes with conditions.
We need to rethink how we measure AI success
AI success is often measured through front-end metrics:
• Conversion rate
• Average order value
• Customer acquisition
These metrics matter. But they only tell us whether AI helped win the customer.
They don’t necessarily tell us whether the business successfully kept them.
There’s an even more basic measurement gap hiding in the data. Retailers expect AI-powered carrier management to cut total delivery costs by 10.6% on average, and more than 60% are banking on savings of 10% to 14%. Yet only 35% can put a clear cost against the main errors and exceptions in their delivery operation today, which leaves 65% working from a partial picture at best.
If you can’t see what your delivery failures cost you now, you’ll struggle to prove the saving later. Better measurement of the operation is what makes the AI’s value provable in the first place.
The next opportunity for AI could be much deeper in the operation.
Predicting stock issues before they affect availability. Identifying fulfilment bottlenecks before they cause delays. Assessing carrier performance and capacity. Proactively communicating with customers when an order is at risk.
This is where AI starts to become more commercially interesting.
Because the opportunity isn’t just to create more revenue at the point of purchase.
It’s also to protect margin, reduce avoidable costs and strengthen the customer relationships that create long-term value.
Where the promise is kept or broken
I’m excited about what AI can do for retail. The potential to improve discovery, personalisation and conversion is significant.
But I also think the industry should be looking beyond the point of purchase.
The customer relationship doesn’t end when an order is placed. The promise has only just been made.
The real opportunity for Agentic Commerce may be helping retailers keep that promise more consistently. Helping them identify risks before they become problems, respond to disruption faster and make better decisions across fulfilment and eCommerce delivery.
And there’s a sensible place to start, where customers already trust it. 64% are comfortable letting AI track a delivery, and real-time tracking is the single improvement they most want (29%). Get last mile delivery tracking right first, and you earn the room to automate the harder calls later.
The future of AI in retail won’t be decided solely by who has the smartest storefront or the most seamless checkout.
It will also be decided in warehouses, fulfilment operations and transport networks. In the moments where promises are either kept or broken.
And from a revenue perspective, that might be the biggest opportunity of all.
Because the real value of AI isn’t just helping retailers win more customers. It’s helping them give those customers a reason to come back.