When the agent checks out: what customer loyalty becomes in agentic commerce

For thirty years, loyalty has been built to persuade a person. Points, tiers, status, the small thrill of an upgrade: all of it assumes a human being is sitting there making a choice. We designed programs to earn that choice, and we got good at it. However, that assumption is now breaking. 

More shoppers are handing the decision to an AI assistant. They ask it to plan the trip, compare the options, and complete the purchase. McKinsey has even forecast that AI agents could mediate $3 trillion to $5 trillion of global consumer commerce by 2030 and become the “first stop” in the shopping journey. When that happens, a loyalty program will no longer be making its case to a customer. It will be read, or skipped, by software that decides in milliseconds what is worth surfacing. The question is no longer only whether a person feels loyal to your brand. It is whether your program is legible to the machine acting on that person’s behalf. 

I have spent more than two decades in marketing and loyalty, and I think this is the most consequential shift the field has faced. It is also the most misunderstood. Many brands are treating agentic commerce as a distant problem. In fact, it is becoming important infrastructure right here and right now. 

The old audience was a person. The new first audience is AI. 

Consider where a loyalty program comes in during an AI-mediated purchase. A traveler asks an assistant to book a hotel for a trip. The assistant weighs price, location, cancellation terms, and availability. If it cannot easily read that the traveler holds points, status, or a redeemable reward with a given brand, that value effectively does not exist in the decision. The customer earned it, and the program holds it, but the agent never sees it. 

This is the part that should get every loyalty leader’s attention. A reward your customer cannot use through the channel they are actually buying in is not a benefit. Instead, it becomes a liability you are still paying to maintain. 

For decades we optimized loyalty for human attention: the right offer, the right message, the right emotional moment. But we now have a second audience that arrives first, and it does not respond to emotion, it responds to structure. It needs data it can read, trust, and act on. If your program cannot provide that, the agent will route around it. 

Structure for the machine, recognition for the human. 

I want to be clear about what I am not saying. I am not saying loyalty stops being human. The opposite is true. The brands that will succeed here will be the ones that treat the agent as the first audience without abandoning the relationship underneath it. The agent is the gatekeeper at the moment of consideration, but the human is still the reason the relationship exists at all and makes final buying decision. You have to serve both, and they need different things. 

The machine needs your loyalty data to be structured, real-time, and interoperable. Balances, status, and redemption rules have to be expressed in a form an agent can parse and act on instantly. It can no longer be buried in an app a customer has to open separately. 

However, humans still need to feel recognized, valued, and understood. Emotional connection remains the thing that turns a customer into an advocate, someone who actively chooses and talks positively about your brand. Our recent research found that 94% of consumers say it matters that brands live up to their promises, yet many brands still fall short. With agentic commerce, that gap could widen or close, depending on whether brands act. 

And the mistake is to treat these as a trade-off when they are not. Structure is what lets the emotional value reach the customer in the moment that now matters most. If the agent cannot see the reward, the customer never feels the recognition. Making the reward legible to the agent is how you protect the human relationship, not how you lose it.  

What loyalty leaders should be doing now. 

This is not a call to rip out the foundation of what you’ve built, but it is a call to make that foundation readable. Start by asking a plain question about your own program: if a customer tried to use their points or status through an AI assistant today, could they? For most brands, the honest answer is no. That gap is the work. 

A few things matter more than the rest. Your loyalty data has to be accessible in real time, not in a batch that updates overnight. It has to be structured so that an external system can interpret it without a custom integration for every assistant. And it has to be governed, because the moment you let agents act on a customer’s behalf, trust, accuracy, and fraud protection stop being back-office concerns and become the product. 

Emerging standards are already making it possible for loyalty data to be shared with AI agents in a consistent way. The brands that adopt those standards early will be present in the conversation, and those that wait will discover their programs have quietly gone invisible at the exact moment customers are deciding what to buy. 

The first move is changing hands. 

For the history of the industry, the first call in a purchase belonged to the customer, and our job was to influence it. Increasingly, agents are taking ownership of that action on behalf of the customer. However, this is not the end of loyalty. It is its evolution, a new audience we have to earn, sitting in front of the one we already know and cater to. 

The brands that prepare for the agent as the first audience, and keep the human relationship flourishing behind it, are the ones that will stay relevant.