Loyalty in the age of AI
New-era loyalty programs will be a critical moat in the agentic future
TGIF! This is my first time publishing on Friday. It’s also my birthday! I’m pretty sure 34 is one of the least exciting birthdays, but at least I can claim I’m in my “early thirties” for one more year. I’m hosting some friends in the Catskills and we’ll do very 30s-ish activities like birdwatching and drinking one glass of wine before falling asleep.
I write a lot about how the e-commerce stack will evolve as shopping becomes agentic. I’ve really enjoyed Scot Wingo’s Retailgentic podcast to keep me up to speed and thinking about this space during my commute (not sponsored - just love it!). The recent episode with Karl Haller from IBM highlighted the prediction that ~25% of e-commerce will be agentic by 2030. This implies ~4% of total retail, as e-commerce penetration has hovered around 20% of total retail for the past few years. While not huge, agentic shopping will become a real (and growing) sliver of how we shop in the near future.
This forecast is in keeping with my Value x Vibes framework - agentic commerce is already starting to penetrate certain categories (like grocery, travel) and will expand from there. Some of these early categories may see significantly more than 4% agentic share by 2030.
This raises a lot of questions about the role of brand. “Agentic” shopping is distinct from getting recommendations on ChatGPT. To be agentic, the tool acts on the consumer’s behalf to make a purchase. Today, agents suggest a few products that fit our parameters and then take guidance from us on those options before purchasing. In future, we can provide more guardrails upfront (or these will be learned), and we’ll need an ever-lighter touch from need to purchase. In this world, brands lose a lot of control.
A few questions on my mind:
What is the role of top-of-funnel marketing in the agentic age?
Will loyalty programs still be relevant, or will we see a race to the bottom on price?
Will we still need brand websites?
How will brands “market” to agents?
Today we’ll talk about loyalty.
Let me know which of these (or others!) are top of mind for you in the comments, so I can cover in the future.
Loyalty programs are everywhere
Loyalty programs aren’t new - they’ve been around since the barter economy. In the village marketplace, the baker might slip the butcher an extra roll with the promise, “Come back next week, and I’ll make it worth your while.” Whether it’s bread, punch cards, or points apps, the core idea has always been the same: reward repeat business and keep customers coming back.
A few key reasons brands maintain loyalty programs:
Build a relationship with customers and drive brand love
Steal share of wallet from competitors and drive incremental purchases
Make the customer relationship “stickier”, and increase switching costs
Gather customer data that can be used to personalize the shopping experience
Loyalty programs buffer brands from needing to compete solely on price and convenience. In NYC, there are coffee shops on every corner. Yet, I choose to walk further to Starbucks and spend more because I love their loyalty program.
Loyalty at risk
In the agentic era, when a shopping agent decides what to buy on your behalf, the parameters driving that choice may never be visible to the consumer. Price and convenience—usually measured in speed of delivery—are almost certain to be among the top inputs.
For less differentiated categories - paper towels, batteries, cleaning products, or plain white T-shirts—brand may simply disappear. The shopper won’t be scrolling through a site, weighing brand names, and adding a familiar product to their cart. They’ll just say, “I need more paper towels,” and the AI will select whatever meets the criteria at the lowest cost and fastest shipping speed.
This shift puts traditional loyalty programs in a precarious position. For decades, these programs have relied on brand visibility and direct customer engagement to drive repeat purchases. If customers no longer visit your website, open your app, or see your promotions in-store, the levers that once powered loyalty—points, perks, and personalized offers—may never make it into the AI’s decision-making framework. In this world, the benefits of loyalty programs could shrink dramatically, especially for brands that live in categories where function outweighs identity.
Future-proofing loyalty programs
Some brands are already leveraging GenAI to make their loyalty programs more compelling. Starbucks - my personal favorite - has started using an AI-powered analytics tool called DeepBrew to personalize offers more precisely than the generic segments many of us are familiar with. This looks like:
A push notification a few minutes before a customer’s usual order time with a special offer on a new drink
A compelling offer aimed at infrequent shoppers to draw them back in-store after a hiatus
A food/drink pairing offer informed by the customer’s order history
These tactics seem to be working. “In the quarter, Starbucks saw its rewards program 90-day active member base in the U.S. increase to 34.3 million, up 13% year over year, amounting to about 4 million new members.”
I’m betting Starbucks is already thinking about preparing their loyalty program for agents - and I look forward to following along.
What does an AI-native loyalty program look like?
The new wave of loyalty programs will be optimized for the agentic era. To do this, loyalty programs need to evolve in 3 key ways.
Show up in agentic workflows
How does it work? Brands must expose their loyalty program data - points balance, perks, tier benefits - through APIs. Shopping agents can pull this data in real time when evaluating purchase options, just like they check price and shipping speed. This ensures that the value of loyalty perks - not just sticker price - will be used to determine an agentic recommendation. In our coffee example, an agent will factor in Starbucks’s free pastry offer when comparing against ordering my daily cold brew from Dunkin.
While still in it’s early days, Loyalty-as-a-Service, or LaaS, will start to show up in more conversations as agentic conversations mature.
Dynamically offer rewards tailored to each agent interaction
How? Building on LaaS, loyalty programs must themselves be agentic - permissioned to offer different rewards in real-time while “negotiating” with a shopper’s agent. This sort of logic will be akin to how many predict that branded agents will offer personalized pricing and promotions. While guardrails must be established, truly personalized offers in real-time will delight the end consumer while converting a sale and building the relationship.
Dunkin will counter Starbucks’s offer of a free pastry with an offer of the customer’s favorite donut.
Use predictive capabilities to pre-empt agentic bartering
What’s better than winning a negotiation? Never needing to negotiate at all. Best-in-class AI-native loyalty programs will pre-empt a consumer ask by pushing recommendations to the shopper’s agent proactively. If I get Starbucks every weekday, why wait for my agent to make a request? Starbucks’s loyalty agent can lock in a week of coffee orders on Sunday.
I’m excited to see early versions of AI-native loyalty. A tough thing about writing about agentic commerce is that many of the capabilities are too early to find real-life examples. While I’m convicted that these shifts are happening already, it can feel “pie in the sky” to talk about capabilities that don’t yet exist. I suspect a company like Starbucks - which has long been a digital leader, and has products which are frequently consumed by a loyal customer base (meaning they have rich customer data profiles) - will be early to the agentic game. I’m sure lots of SaaS startups are also jumping on this idea. Time will tell!
If you’ve seen anything like this, or if you’re building in this space, please reach out - I would love to chat!
Stay Curious,
Melina
What a great post, very well thought out. As an ecommerce advisor to many brands, loyalty is such an important part of retention; it will be interesting to see what happens in the coming months and years. I'm interested to see how it plays out with Shopify's Agentic Commerce initiative and their new universal checkout. I suppose loyalty will have to become universal as well.
Loved this - and had been thinking about this a bit as well.
- Maybe the agent itself has a loyalty program. People may be choosing between agents - maybe the loyalty is not to the retailer but instead to the agent?
- Could the unlock be for the choices of retailers that the agents might be "visiting"? Maybe some of the retail choices are more welcoming to agents - is their data organized in a way that results in more customer satisfaction, less returns, etc? For apparel, are their product descriptions more detailed resulting in a higher level of success?
- If you use an agent service that is not associated with a specific retailer, could you direct your agent with guardrails to stick to certain retail destinations? If, for example, a reward is experiential and meaningful, then the human would give the agent more direction?
I guess the biggest question I have is where the loyalty will be - with a retailer? with a brand? with an agent?