Marton Kosdi-Kovacs: Beyond points. A deep dive into e-commerce loyalty programs

Intro

Matt: Marton, for anyone who doesn't know you – a quick introduction. For regular subscribers, we've covered the basics of loyalty programs in a previous episode. Today we're going deeper.

Marton: Thanks for having me. I'm Marton, building Love Loyalty with two co-founders and a team of eight. Love Loyalty is a retention platform for brands on Shopify – offering points programs, VIP tiers, referrals, paid memberships, and everything you need to run an engaging, well-functioning loyalty program. Both for online Shopify stores and for stores with physical locations.

AI in development – over 90% of code written by machines

Marton: We're a team of eight, hiring mainly for support and customer success because we do a lot of onboardings. On the development side we have one or two people plus our CTO – but over 90% of the code we're shipping today is written by AI.

Matt: I've heard similar figures from other brands. The interesting part is why they're not firing their developers.

Marton: Our CTO built a tool where one AI model writes the code and another reviews it – through about ten rounds of review. A human still looks at the final output before it goes to production. For small fixes and minor features, sometimes he doesn't even need to step in. But AI doesn't understand the product, the strategy, the roadmap. You still need a human who understands what you're building.

Matt: Completely agree – a human in the loop is essential. You also need guardrails so agents operate within a limited scope. We were actually discussing this on our Slack today – people flagging quality issues with the new Shopify partner dashboard, and the likely suspect is AI-assisted engineering on less critical parts. I think it's temporary. Once the workflow is solid, quality will catch up.

Marton: I'm genuinely happy that Shopify is leaning so hard into AI. We're actually launching our Sidekick integration now, so merchants will be able to configure loyalty program rules directly through a chat with Sidekick.

How to measure whether a loyalty program works – attribution and metrics

Matt: Let's start with the maths, because this is the most common question from merchants: how do I know if my loyalty program is actually generating revenue? Would those orders have happened anyway?

Marton: Attribution is one of the hardest questions in loyalty. But I think brands need to treat launching a loyalty program as a must-have when they want to maximise retention. No one will ever tell you with certainty that one specific purchase happened because of the program. But if you launch, it resonates, and it works – you'll see it in the numbers. First-time buyers come back more often. Your most active customers have higher AOV.

What we recommend before you launch: measure your baseline. What percentage of your customers are returning? How often do they come back? How many first-time buyers return? Then launch, iterate – very few brands get it right on the first attempt. After twelve months, compare those same metrics for the same customer group.

And look at point redemption – it's the simplest indicator. If customers who redeem points buy more frequently and spend more than those who don't engage with the program, it's working.

Matt: Agreed. These trends only become clear over a year or more – in the short term too many external factors overlap. I like LTV segmented by loyalty activity as a metric. But ultimately everything depends on having clean, trustworthy data. That's where the Sidekick integration will help – you'll be able to ask difficult questions in plain English and get the reasoning done for you.

Marton: Exactly. We're doing a lot of migrations from legacy loyalty solutions and when we ask merchants how their loyalty is performing – most have no idea. They're paying $600 a month for a tool and have no clue whether it lifts their metrics. E-commerce managers need to know their target numbers and actively track purchase frequency and LTV.

Margin baseline – how much can you give away before hurting profitability

Matt: How do you figure out the discount threshold where your loyalty program stops being profitable?

Marton: It depends on your margin. We have clients with 80% margins – for them a 10% discount is no problem at all. We also work with subscription brands that only reach profitability after the fourth purchase. You have to know your margins and decide what discount is worth offering and what discount your customers will actually care about.

The benchmark we recommend for most brands: after the first purchase, award points worth between 5 and 10% of the next purchase. That's enough to do something in the customer's mind – they think about coming back. If they need another pair of shoes and they know they have a 10% discount waiting, they'll probably go back to you.

Non-financial rewards – access, status and belonging

Matt: I wanted to talk about non-monetary rewards. Strategies that are perceived as highly valuable by customers but relatively cheap to implement.

Marton: This is a growing direction. Purely transactional programs – one point per pound spent – have become the standard. They no longer differentiate anyone. I think the future of loyalty programs is not transactional at all. These programs should generate a feeling of access, belonging and status.

What we see working:

• Invite-only events – give a genuine sense of status.
• ‍Early access to new collections – particularly effective in fashion and beauty.
• ‍Closed communities with product feedback – a customer reaches a certain tier and gets access to a space where they see unreleased products and can give feedback. They feel the brand genuinely listens and that they belong there.

‍One-on-one meetings with the founder – sounds extreme, but some brands do it and it works.

Matt: Glossier runs focus sessions with their top customers – literally asking them what's missing from the offer, what isn't working. Win-win: the brand gets insights, the customer feels they have a voice.

Marton: Exactly. A closed community gives brands one of their most valuable assets – regular, honest feedback from people who genuinely care.

Loyalty program as a data engine

Matt: Something that's rarely discussed – the loyalty program as a data collection tool, not just a revenue driver.

Marton: This is one of the most underrated reasons to have a program. Customers sign up giving you their email, name, birthday. That immediately unlocks birthday campaigns, personalised email marketing, product recommendations on their next visit.

The more information customers willingly share – because they want points and a better experience – the more personalised your communication can be. Take pet food: ask about the dog's breed, how often it's walked, what it eats. You can then recommend exactly what that specific dog needs. The customer is happy to share because they care deeply. For beauty – skin type, ingredient reactions, seasonal recommendations. For apparel – fit and style preferences.

Matt: And at the most basic level: if you know someone has a dog, you don't send them a campaign about cat food.

Omnichannel – connecting online and in-store into one program

Matt: What about customers from physical stores? How do you bridge offline and online into a unified experience?

Marton: This is a very hot topic right now. When a customer walks into a physical store and leaves, the brand usually has zero information about them. We estimate that 70 to 80% of in-store customers are unreachable after that first purchase. That's a huge loss.

Shopify is investing heavily in omnichannel and POS – and we're seeing more and more large brands that previously didn't use Shopify POS switching to it specifically to integrate their loyalty program across online and offline.

In practice: all you need is to ask for an email at the till or put out a QR code. The customer scans, joins the program. They can immediately start earning points, receive personalised offers, redeem rewards on their next visit – online or in-store.

At Love Loyalty we're investing heavily in wallet passes right now, because they reach customers who don't read emails. A branded t-shirt at £100 is often bought by a 40–50 year old who doesn't have time for email campaigns. Wallet pass notifications convert significantly higher than email.

We also have a near-store notification feature – if a customer is within a mile of your physical store and has accepted location sharing when adding the wallet pass, they can receive a notification: "You're nearby – come in today and get 10% off our new collection."

The staff script – two sentences that decide in-store onboarding

Marton: We work with a brand called Felix & Norton from Canada who run a paid membership program. They experimented extensively with what staff say to new customers at the till. It came down to the exact two or three sentences the salesperson uses – not the discount level, not the points value. Get those sentences right and the customer joins. Get them wrong and you've lost them.

Matt: An underrated detail. You can have a great program, but if the staff don't know how to present it in that one moment, you lose 80% of the opportunity at first contact.

Quiet periods – using your loyalty program to address seasonality

Matt: Most industries have quiet seasons. Can a loyalty program help fill those gaps?

Marton: Yes, and we have a few strategies that work well.

Expiring points – set points to expire during your quiet period and send campaigns to those customer segments: "Your points are expiring soon – use them now." Works very effectively.

Tier progress – notify customers that they're close to the next tier. "Make a purchase now and you'll unlock level X with these rewards."

Double or triple points – add bonus points during the quiet period, optionally for specific products or collections. If a haircare product doesn't move between June and August, run triple points on it for exactly that window.

As a side benefit: these same mechanics help with inventory management. You can move slow-moving stock without opening a sale.

Red flags – when your loyalty program is hurting more than helping

Matt: What signals should tell a brand that their program is giving too much away?

Marton: Three main red flags:

• Redemption rate too high – a healthy program runs at 20 to 40% point redemption rate. If you're consistently above 40% month after month and profits aren't growing proportionally, you have a problem with your program rules.‍
• Discounting customers who would have bought anyway – if after launching the program you see customers only coming back to redeem points, never buying without a discount, you're burning margin on people who would have converted regardless.‍
• Reward cost exceeding retention value – you need to calculate: is the total cost of the program plus the value of rewards given away lower than the incremental revenue from retention? If not, the program isn't paying for itself.

Paid memberships – when they make sense and how to introduce them

Marton: Paid memberships work very well – but not for every brand. We see paid members spending three times more and buying three times more frequently than returning customers without a subscription. It's a psychological effect: "I paid for this – I need to use it."

When does a paid membership make sense? When at least 30 to 40% of your customers are returning customers. That's the signal that you have a base of people who naturally come back – and you can convert them into paying members.

One important caveat: the brand needs a level of recognition first. Nobody is paying even $10 a year for loyalty to a brand they've never heard of.

For in-store rollout: the best moment is the first purchase. If a product costs £100 and an annual membership is £20 with an immediate 15% discount, the customer walks out ahead – and the brand gains a paying member. The staff script is critical here, same as with any first contact.

AI loyalty builder – a complete program designed in minutes

Matt: I know you've launched something new that addresses one of the hardest parts of getting started – actually designing the program.

Marton: We've just launched our AI loyalty program builder for new merchants installing Love Loyalty. You tell it your goals, it checks your orders, products and customers, and in a few minutes designs a complete set of earning rules and reward types – including a suggestion for what discount level makes sense given your margins. You review it and launch.

This addresses a real problem. We've seen brands spend months designing a program, and most of them end up copying a competitor and offering slightly more points. That's the wrong approach – if you're identical to everyone else, customers have no reason to be loyal specifically to you.

Matt: I tested Sidekick today on loyalty program design. Interesting results – to be truly effective it needs cost of goods (COGS) entered for products so it can calculate gross margins and make meaningful recommendations. Once Love Loyalty connects to Sidekick, that combination will be genuinely useful for merchants.

Marton: Sidekick is getting smarter – and I think it will become a must-have for modern loyalty programs. In loyalty, data and the interpretation of data is everything.

Loyalty is not a set-and-forget task

Matt: I see a lot of merchants treat loyalty as a one-time task: set it up, tick it off, move on. Is that a viable approach?

Marton: Not at all. I think part of it is cultural – in Europe the cost of customer acquisition is still relatively low compared to the US, so retention doesn't hurt as much yet. That will change.

A loyalty program requires iteration. Very few brands hit the optimal configuration on the first attempt. You need to measure, adjust, test. In twelve months I'd like merchants to be able to say: "I tend to this program, I experiment – and I can see it's affecting my revenue." That's the goal.

Final thought – start by knowing your customer

Matt: Last word – what would you say to brands that are only just thinking about a loyalty program?

Marton: The same thing I said last time, because it's still true: every brand can have a loyalty program, no matter what you sell. The tools available today let you build any kind of program – you don't need to limit yourself to the point-per-pound model.

But for that to work, you need to genuinely know your customers. Understand what motivates them, what they're looking for, what will make them stay. Understand your customers – everything else follows from that.

Matt: Marton, thanks for the deep dive. See you at the next one.

Marton: Thanks for having me. Until next time.