Is AI getting in the way of your relationships?
Let me walk you through a purchase process that is becoming ordinary, and show you where your loyalty programme either shows up or completely disappears, writes Denise Holt, SVP, Head of Strategy, Experience, Research + Insights, Phaedon
A customer needs a hotel for a summer vacation. A year ago, she would have opened an app, searched, scrolled, weighed a few options and booked. Today she types a sentence into an AI assistant: find me something near the seaside, good cancellation terms, under a certain price, and sort out the booking. Then she goes back to her day until the assistant responds. A recent NYU SPS and BCG analysis found that 37% of travelers already use AI large language models embedded in online travel sites to plan and book trips, underscoring how quickly the booking journey is shifting from search and scroll to ask and book.
Follow her request through each stage, and one extremely important question confronts every loyalty leader: when the assistant is doing the choosing, does your programme show up or influence results? At every point where your programme used to reach her directly, it is now reaching the assistant acting on her behalf. Or it is not showing up at all. The good news is that this is becoming winnable, and it asks something major that teams can work to deliver on.
Stage one: the AI assistant draws up the shortlist
At this first stage, the assistant is comparing options on what it can read: price, location, cancellation terms, availability. This is the first moment that matters for your programme. Say this customer has points or elite status with your hotel brand, earned on past personal stays. If the assistant cannot see that, it counts for nothing in the shortlist. She earned it. Your programme is still holding it. But the assistant simply never sees it.
For decades, we earned a place on the shortlist through human attention by delivering the right message at the right moment. The assistant does not respond to a clever message. It responds to data it can read and act on.
So, the first piece of work is plain: your loyalty value has to be legible to the systems doing the comparing and integrate with AI engines. Real-time balances, status and redemption rules, expressed in a form an external assistant can interpret instantly, rather than locked inside an app or inbox she now rarely opens. This is what being considered now requires.
Stage two: she tells the assistant what good looks like and narrows down the list
Now the human preferences really enter, and this is where the relationship can be won or lost. She does not just want the cheapest room. She has preferences, a brand she trusts, and a sense of who has treated her well. She expresses that to the assistant in plain language, and a good assistant weighs her instruction heavily. This is the opening for loyalty and brand recognition, and it is enormous, because a customer who actively prefers your brand will steer her agent toward you.
The catch is that the preference has to be real. Our latest research into emotional loyalty points to a persistent execution gap: customers rate these emotional drivers as highly important, but far fewer feel the brands they use most deliver on them. To take one example, 86% of consumers say it matters that brands act on their feedback, yet only 67% say the brands they use most do. The rewards are rarely the issue. Feeling heard, recognised and understood is.
That preference also varies by generation, which is why a single averaged view of the customer leaves value on the table. In our research, 89% of Baby Boomers say it matters that a brand goes above and beyond, yet only 69% say the brands they use most do. Among Gen X, 94% say it matters that programmes provide flexibility in how they redeem, and only 76% say the programmes they use most do. For Gen Z, 84% say it matters that a brand treats its staff well, and only 71% say the brands they use most do. Earn the preference on the terms that matter to each specific generation, and you are the brand she names when the assistant asks what she cares about.
Stage three: the booking happens, or it doesn’t
If your loyalty value was legible at stage two and her preference for you was genuine, this is where it pays off: the reward is applied, the status recognised, the booking made manually by her, all inside a few exchanges she barely had to think about. The programme delivered in between these moments that now decides everything.
If either piece was missing, the booking still happens, just not with you. And she is none the wiser. There is no abandoned basket to chase, no lapsed-member email that lands, because from her side nothing went wrong. She asked, the assistant delivered, a brand with the best AI visibility won.
What this asks of us
Two things, and they work together rather than against each other. First, make the programme readable by the machine: loyalty data that is real-time, structured so an assistant can use it without a bespoke integration, and governed, because once agents act on a customer’s behalf, accuracy, trust and fraud protection become part of what you deliver rather than back-office detail. Emerging standards, including the Model Context Protocol, already make it possible to share loyalty data with AI agents consistently, and the brands adopting them early will be present in the conversation.
Second, keep earning the human preference that tells the assistant to choose your brand by closing the gaps our research identifies: act on feedback, recognise people as individuals rather than segments, and get the generational nuance right. Legibility gets you onto the shortlist. The relationship is what wins the booking once you are there. You need both, and the brands that pair them will find that the assistant becomes an ally rather than a gatekeeper.
About the author: Denise Holt, SVP of Strategy and Growth, Phaedon
With over 25 years of marketing experience in loyalty, brand strategy, and consumer insights, Denise leads Phaedon’s strategy, experience, and research team. She has spearheaded industry-leading research initiatives, including “Humanizing Loyalty” and “Sparking Consumer Participation Through Shared Values and Beliefs.”