Marton Kosdi-Kovacs: Więcej niż punkty. Analiza programów lojalnościowych w e-commerce

Intro

Matt: Marton, for anyone who doesn't know you, could you give a brief introduction of yourself and the software you represent? Regular subscribers will remember Marton. We recorded an episode before and covered some basics, and Marton showed us some of the ropes of loyalty programs in e-commerce. Today we're going to do a deep dive. But let's start with a quick introduction.

Marton: Thanks for having me. I'm Marton. I'm building Love Loyalty along with two co-founders and a team of eight now. Love Loyalty is a retention platform for brands on Shopify. It offers points programs, VIP tiers, referrals, and paid memberships, basically everything you need to set up an engaging loyalty program, both for online Shopify stores and for stores that also have a physical location.

Matt: You said you're a team of eight now, right?

Marton: Yes, though it's constantly changing because we're hiring more people for support. We do onboardings and setups for brands, so we need support colleagues and customer success managers. On the development side, besides our CTO, we have one or two people, because right now, in this age of AI, more than 90% of the code we're shipping is done by AI.

Matt: I've heard that from many brands. I think the guy from Spotify claims they have 100% of their code shipped by AI. The interesting part is why he's not firing his developers, right? But I guess that's the new normal.

Marton: Exactly, tokens. Our CTO built a tool using several AIs, where one AI builds something and another checks and reviews it, with around 10 rounds of review. He looks at the finished version before it goes into production. But for small fixes, or features that are just adding a button somewhere, it doesn't even really need review, it just works. Still, you obviously need a human there, because someone needs to understand how the program works. It has to have a human touch, in my opinion, and that's why we have another developer who understands the product. AI won't understand what we want to build, what our strategy is, or what our roadmap looks like. It just builds.

Matt: I totally agree that you need a human in the loop, someone responsible for quality. The other thing is you need guardrails for the agents to work within, giving them access to a limited scope of functionality. We explore this all the time. We even had a conversation earlier today on our Slack channel where people were complaining about the quality of the new Shopify partner dashboard, where many things don't seem to work properly. The suspect in that puzzle is obviously AI agents coding, or vibe coding, some of the less critical functionality. So AI-assisted engineering is taking its toll. But I think it's temporary, and once the workflow is solid, it'll be top-notch quality.

Marton: Of course, it'll still take time. But I understand what you're talking about, because we experience the same as app developers on Shopify. I'm really happy that Shopify relies a lot on AI. For example, there's Sidekick now, and merchants can build a few things on their own. We're also launching our Sidekick integration, so it'll be even easier to launch a new reward type or rules for your loyalty program, because you can just chat with Sidekick.

Matt: That recent update from Shopify, where Sidekick can integrate and interact with apps from the ecosystem, is maybe not a game changer, but it's a nice feature. We see brands moving toward this AI operation mode when running their e-commerce business, right? We'll get back to Sidekick, because I ran some tests on Sidekick and loyalty today, and I'll tell you more about that.

Marton: Okay.

The math behind loyalty: how do you know it's working?

Matt: Let's jump back to the main subject of today's episode: loyalty. Over the last few years, we've had more and more conversations with our customers and prospects about loyalty. I've collected some of the most common questions and objections that come up, and I'd like you to help me elaborate on them or handle the objections, and give people arguments or a way to understand what they can actually do with their loyalty program. I'd like to start with the math behind loyalty, because that's probably the main concern: how do I know if my loyalty program is driving revenue? How can I tell that these orders come from the loyalty program, or that they wouldn't have happened anyway?

Marton: This attribution question is a big part of the conversation, and it's one of the hardest questions in loyalty. Brands need to look at building or launching a loyalty program as a must when they want to maximize retention. If you implement one and it works, you'll see that it's working. No one will tell you that one exact purchase happened just because of the program. But if it starts to resonate with your customers, and you see first-time buyers coming back, or your most active customers coming back more frequently or with a higher average order value, it means your program is working.

There are ways to measure it, so you don't have to rely on gut feeling. What we usually recommend is that before you launch, you understand all your retention metrics: what percentage of your customers are returning, how often they return in a specific period, how many of your first-time customers come back, and so on. One reason we ask this is that it tells us what type of customers you have and what type of program to design. The other reason is that it's good for analytics after launch. If you look at a group of customers over the six months before you had a loyalty program and understand their retention and spending, then launch and iterate on your program (because very few brands get it right the first time), you can look at the same group again in 12 months. Compare their spending behavior: how much more frequently they purchase, how much more they spend when they come back, how they use their points, how active they are in the community. Put all of that together and the numbers will tell you whether your program is working. Another thing that correlates is point redemption. You analyze whether people redeeming points spend more and come back more often than customers who aren't active loyalty members. You can do the same with average order value.

Matt: I totally agree. My take is that you need to evaluate it over the long term. A quarter probably isn't enough. These trends emerge when you compare year over year. In the short term, launching a loyalty program can coincide with other factors, so it's hard to attribute. I like average order value, or lifetime value if you use segments, as a clear indicator. But at the end of the day, it comes down to having your data right and trusting it. That's critical. Especially with this new Sidekick integration, it might be easier, because you can ask these difficult questions in plain English and it'll do the heavy math and reasoning for you.

Marton: Exactly. There's no Excel sheet where you enter your order numbers and it tells you how much retention you should have after two months, because it's different for every type of product and every brand. The problem, and this is why it's a great topic, is that we do a lot of migrations from large or legacy loyalty solutions. Whenever we ask brands how their loyalty is going, or whether it's working, they have no idea. They pay maybe $600 a month for a tool and have no idea whether it lifts their metrics. They have no idea about point redemption. I tell them they should at least measure these things. That's why Sidekick will be a great help. E-commerce managers should learn more about the target numbers they should have for their loyalty program, and measure LTV and purchase frequency. You're right that it needs more time, six to 12 months, until you see whether a program is really working. But even after one or two months, if you just look at point redemption, which is a super simple metric...

Matt: Of course.

Marton: ...you can tell whether it works or whether you have to adjust it.

Matt: If the simple question is "is it working," then I totally agree. But calculating profitability might be trickier, right?

Marton: Yeah.

The margin baseline: how much discount is too much?

Matt: The next question is about the margin baseline. What would you say is the threshold of discounts that stops your e-commerce from being profitable, where the discounts or the points-awarding system harm your profitability? How do you wrap your mind around that?

Marton: Again, it's different for each industry and each brand. We see companies using our loyalty program with 80% profit margins, and they can give away a 10% discount without any trouble. There are also subscription brands that don't make money on the first purchase, or even the second. I heard of a huge brand, not one of our users, that only reaches profitability after the fourth purchase. So you have to know your margins and your customers, and based on that you can decide what discounts you can give away and which ones customers will actually care about. As a benchmark, what we usually recommend, and what works for most brands and motivates customers enough, is that after one purchase you give points worth around 5 to 10% of the next purchase.

Matt: Between 5 and 10%.

Marton: It does something to the customer's mind. They start thinking about it and might want to come back. If they need another pair of shoes and you have a shoe store, and they know they have 10% worth of discounts there, they're probably going to go there. But it's not always the discount that works. If it is about the discount, you have to calculate the margin on the product you're discounting and how that discount increases your retention overall.

Non-financial rewards: access, belonging, and status

Matt: You touched on something I really wanted to talk about: non-financial gains that brands can offer in their loyalty programs. This is interesting because I've heard about strategies that are perceived as very valuable by the customer but aren't expensive to implement. Can you share some of your experience and what you recommend?

Marton: It's more and more popular to have different types of rewards than just the transactional "one point per dollar spent." If you look at e-commerce, it was maybe a bit more than 10 years ago when loyalty programs became popular, and everything was transactional. After a while, all brands had these programs, even today. If it's the same for everyone, customers won't care which one they buy from. Your goal is to make more money and increase retention, and for that you need brand loyalty, people who actually want to come back to your store. That's the future of loyalty programs: not staying transactional, but generating a feeling of access, belonging, and status.

That's why we push brands to first understand their customers. Based on that, they'll know what matters to them. Larger brands come to us saying exactly that: "We don't want a simple points program, we want something different. Can you show us a few beauty or cosmetics brands doing something more than a transactional points program?" That's why we put a lot of energy into building custom solutions, and into giving e-commerce managers more tools to build their own unique loyalty experiences.

Matt: I totally agree. If your only strategy is giving away discounts, you don't really have loyalty. You don't have brand ambassadors or advocates, you have point collectors and discount seekers. The psychology behind loyalty is an interesting conversation, because people don't always seek transactional rewards. They seek status and things that are perceived as available only to them. I've heard about brands doing queue skips: if you reach a certain loyalty tier, you're served first. It doesn't cost the merchant much, but customers perceive it as a signal of status. The same goes for express returns. I did some research today and found a crazy example: some fashion brands experiment with home try-ons, where a courier delivers clothes to your door and waits for half an hour while you try everything on, and takes back whatever you don't want.

Marton: Yeah.

Matt: So you can come up with anything and build it into your loyalty program, and test it fast.

Marton: One of the things getting popular is member-only or invite-only events. That really gives a feeling of status. Then there's access to specific collections, and here I'm mainly talking about apparel, fashion, beauty, and accessories brands. For a pet food company it's a bit different in terms of what unique experience you can provide. But invite-only events, access to specific products, or early access to products all work. There's also some insane stuff. For example, if you reach a certain tier, you get invited to a community where you see products that haven't launched yet and give feedback, and based on that the company can take action. If enough people say they don't want a certain style, or they want a certain color, the brand listens. That makes customers feel the brand cares about them and that they belong. Some brands even offer one-on-one meetings with the founder. It's interesting to see loyalty going in this direction. That said, points programs also work well for some brands, as long as they're well designed with good reward types and free products people actually care about. Points programs can go crazy too.

Matt: A short digression: have you heard about these "dopamine buying" sites, this trend from South Korea? It popped up yesterday, I saw an article on Mashable or somewhere. Bear with me, I'm going somewhere with this. These are fake e-commerce or food delivery sites where young people in Korea pretend to buy things. They shop, add to cart, click buy, and it simulates a transaction. They receive all the transaction emails, they can even track delivery, but they're not paying real money and they're not getting anything. It's just for the experience. There are different theories about why this is happening, but one relates to status, which is what we've been discussing. There's economic turmoil and young people losing jobs in Korea, and the hypothesis is they're so addicted to buying that they still get their dopamine from a fake buying experience.

Marton: I can imagine it, but it's weird and hard to comprehend. But it's happening.

Matt: It tells you something about where we are as a society, and about our buying habits and the need for status.

Marton: Yeah, I get it. You do the shopping for the experience and the status. The dopamine hit doesn't come when the goods arrive, it comes when you click buy. That's insane.

Matt: It makes you think about the psychology behind purchasing. But going back to loyalty and non-transactional rewards: I agree. Brands should see these events or meetings with customers as a win-win, because you get access to insights. I know Glossier does focus sessions with their top customers, who tell them what they'd like to see in the store, what's missing, and what the problems are with current products. As I said, it's win-win for both sides.

Marton: I totally agree. That's why I like the idea of a closed community you reach after spending a certain amount, where you can give feedback on products. It's super insightful for the brands as well.

Loyalty programs as a data-harvesting engine

Matt: Another question, Marton. You mentioned that when loyalty programs first emerged, they were basically about points or tiers. When most people think about a loyalty program, their mind naturally gravitates to that narrative. But you can think about loyalty programs differently. You can use them as a data-harvesting engine, for purposes other than just generating revenue, right? Can you tell us more about that?

Marton: For sure. One of the main reasons merchants want a loyalty program, and can be satisfied with it even without the points side working well, is that they get a lot of information about their customers. Customers sign up with their email, name, and usually their birthday, and brands can use that for different email campaigns. If a customer comes back to the site, and people usually stay logged in, the brand can show them the exact products they saw before. If they gave their birthday, the brand can send birthday rewards or offers. So email collection is an important part of a loyalty program. Any additional information, usually birthdays, an Instagram account, or a phone number, can be useful too. This helps brands bring customers back and retain them.

Matt: I totally agree. Pet owners especially have great potential as a source of insights, because of all the details and hobbies their pets have, which they love to talk about. You can get the best intel and all the details about their puppies and cats, and then target them with very specific, narrowed-down campaigns.

Marton: Totally. Pet owners care a lot about their pets, which means they want to give you more information so you can recommend the best food or anything else. They care as if the pet were their child. The more information you get when signing them up, the better the experience will be for them. If you ask about their dog's breed, how many walks they take per week, what food they're getting, and how often, the brand can get creative and give a super personalized experience, both in email campaigns and when the customer returns to the store and you recommend products. Suddenly the discounts are on the products they really want. The more information you get, the more loyal they can be, because you can use it to create a personalized experience. This isn't only true for pet food. It's also true for beauty brands: they know your skin type, so they know that with your skin type this weather in June calls for a particular cream. They can give you access to videos based on how your skin reacts to things. The same goes for apparel: if you like short or baggy jeans, they'll recommend those and send relevant offers. So a key thing about a loyalty program is that you get more and more information that customers give you willingly. They want to give it so they can earn more points or rewards and get a more personalized experience, and all of that helps build brand loyalty.

Matt: Totally. Even at the simplest level, when you have information that someone has a dog, you don't send them email campaigns about cat food, although my Labrador wouldn't mind eating that either.

Marton: Yeah.

Bridging offline and online: POS, wallet passes, and staff scripts

Matt: So, Marton, what about in-store customers? If someone has a brick-and-mortar store and a POS system, is there a way to gather information about those customers, engage them in the loyalty program, or merge the online and offline worlds and increase their activity?

Marton: Definitely. That's a hot topic right now. When you go to a physical store and buy a shirt, most of the time you just walk out, and the merchant gets no information from you. Brands are really missing out on those customers, around 70 to 80% on average, who won't hear about the brand after that purchase. With Shopify's omnichannel experience, Shopify is pushing hard on retail locations and POS. We see more and more large brands using their loyalty program with Love Loyalty online, who didn't have Shopify POS before but are now switching to Shopify and want the loyalty program in POS as well. When a customer walks in, you can ask them at the counter for their email and have them sign up. You don't even have to ask for the email, you can put out a QR code, they scan it, and they're in the program. That way you know what they purchased and can send relevant offers, give them points, or give them discounts that make them want to come back. For memberships, you can even give them a discount right at their first purchase in the store. It's also useful for people who already purchased online. If your staff asks whether they're part of the loyalty program, they can easily redeem their points right at checkout in the physical store.

At Love Loyalty we're pushing more and more features to POS because it works really well. This unified omnichannel experience, with points, tiers, or memberships both online and in store, is huge. We're doing a lot of wallet pass features this summer. We visited a few of our clients in London, and the wallet pass features really resonated with them. Many busy people look at email campaigns less than younger people do. If someone sells, say, £100 t-shirts to busy 40- or 50-year-old professionals, those customers aren't as keen to look at email campaigns. But if they make a purchase in your store and you get them onto your wallet pass with a QR code, or they add it online, you can send wallet pass notifications, which convert much higher than email.

Matt: Interesting.

Marton: So you can send a notification saying, "You got 200 extra points, use them by the end of June and come back to our physical or online store." Another feature we recently launched is the near-store notification. Brands really like it. If a customer is walking within a mile of your store, you can send a notification to their wallet pass saying, "Come in today and get 10% off our new t-shirt," or something like that.

Matt: What's the technology underneath that? How can you tell the proximity of people to your store? Do you use location data?

Marton: They have to accept location sharing when they sign up to the program and add the wallet pass. It's an additional feature that works for some and not for others. Having these wallet passes is also better because when you're in a physical store and the staff checks your QR code, you can see your point balance and everything in the wallet pass. It's compact, and you don't have to go to the website and check your profile for your points or reward types. It's all in one small wallet pass, so it works amazingly well.

Going back to what I mentioned earlier about first-time buyers, how around 80% just walk out, it all depends on what the staff says and how they ask. We have a brand called Felix and Norton from Canada. They experimented a lot with loyalty and memberships, and it all came down to the script the staff uses. It's just two sentences, but if you get it right, you'll get a lot of your customers onto your loyalty program, and they'll come back. In their case it's a paid yearly membership, and they have a two- or three-sentence script they tell every new customer, and it works really well.

Matt: That's a really good insight, because it's low effort but tricky. You can tell your staff to mention the loyalty program, but the way they do it is essential.

Marton: Exactly. We now recommend these things based on insights about what the staff should say to customers, because there are a few points that must not be missed, and it matters how you tell them. If the staff knows it, it works.

Matt: Now I'm wondering what the winning sentence is, but I guess it's different for everyone.

Marton: Yeah, I guess it's different for every brand, culture, and industry.

Using loyalty to drive sales in quiet periods

Matt: Okay, I have a few more questions. The next one is about using loyalty programs to make people buy in so-called quiet periods. There are seasonalities in some industries. Not everyone faces that, but most industries have some quiet seasons. Can we leverage a loyalty program to make people buy, or redeem their points, during those periods? What's your take?

Marton: For sure. Every brand has quiet periods, and again it starts with analyzing your customers and their purchase behavior. You have to know how many customers come back from day one to day 30, day 30 to day 60, and so on. Then, if you break that down by customer segment and product category, you can find the really quiet periods and adjust your loyalty program to them. One big strategy that works is setting points to expire at a certain time, and setting up email campaigns telling those customer segments that their points will expire soon and they should redeem now. The same goes for tier progress: you can tell customers, "Come back now and you'll reach the next tier, which gives you these rewards." These two strategies really work to get people coming back in slow periods. Others double or triple points during quiet periods, and you can do that for specific products or collections. Say months six to eight have no returning customers for your hair care product. You can target just that product with double or triple points in that period.

Matt: That's smart.

Marton: And it actually helps you manage warehouse issues.

Matt: Right, like a slow-moving product.

Marton: Exactly, it helps with inventory. These are quite simple strategies, and we see them working.

Red flags: are you giving away your margins too fast?

Matt: Marton, I have two more questions before we go. One that comes up a lot is about figuring out whether you're giving away your margins too quickly. Can you point to any red flags that a loyalty program isn't working as it should?

Marton: You mean giving away too much discount and not enough profit? As we discussed, you have to know your margins and what discount is still worth it. What we see is that a really good point redemption rate is between 20 and 30%, or even up to 40%. But if you have a constant month-by-month redemption rate above 40% while your profits aren't growing much, you probably have a problem with your loyalty steps and earning rules. Another red flag: if you measured your customers before launching, and the only ones coming back are redeeming points and using the program, that means you're giving discounts to people who would have purchased anyway. You're just burning money. If you know your margin, you have to measure whether your reward cost is higher than the retention it generates.

Matt: That came to mind when you mentioned asking customers at the counter whether they want to join, and handing them a discount right away. They're already buying, so you could hand them a discount for the next purchase.

Marton: I mentioned that specifically in the context of a paid membership, not a simple points program. For example, if you have a $20-per-year paid membership and you upsell it at the counter during the first purchase, you can give the discount right away.

Matt: Okay, my bad. That makes a lot of sense, incentivizing them and converting them on the spot, because the membership is probably worth less than the discount.

Paid memberships: the psychology and the threshold

Marton: Exactly. We see that brands don't really make money on the membership itself. It's extra revenue, but not that relevant. It's the psychology behind it: if the customer has it in the back of their mind that they paid to become a member, they'll go back and stay loyal, not because they're loyal to the brand, but because they paid and want to use the discount, otherwise they paid for nothing. So they need to keep purchasing. We see paid members spending three times as much, or coming back three times more frequently, than customers who return but aren't paid members. It works really well. To launch a successful paid membership, you still need a bit of brand loyalty and people knowing your brand, because for a no-name brand, I wouldn't pay even $10 a year. But once you reach a level of popularity, paid memberships work really well, both in physical and online stores.

Matt: What would you say is the threshold, maybe not in headcount but in volume or size, where a paid membership makes sense? Is there a rule?

Marton: I usually recommend paid memberships if 30 or 40% of your customers are returning customers.

Matt: I thought it might have to do with the scale of the business, like revenue.

Marton: I don't think so. If it's a type of product people buy more than twice a year and 30 to 40% of your customers return, it means that when they think about it, they come back to you. You can make them even more loyal, or get them to spend more or come back more frequently. So it comes down to the percentage of returning customers.

The AI loyalty program builder

Matt: Okay. Marton, one final question. I know you're working on an extra feature that solves one of the most difficult problems of loyalty programs: setting up the rules and onboarding. Can you share a bit about that?

Marton: Exactly. At Love Loyalty we've already launched our AI loyalty program builder, currently for new merchants who are installing our app. It works easily. When you install Love Loyalty, one option is to tell our AI program builder your goal or the type of program you want. It then checks your orders, products, and customers, and based on all of that it designs a complete set of earning rules and a complete reward system in a few minutes. It also designs what we talked about earlier, how many points or discounts you should give away. It gives you simple reward and earning rules, so you can give 10% back, for example. Then you just review it and launch. It works well, and I'm seeing more and more brands use it. This setup is good for brands who have no idea how to set up their loyalty program, or who want to quickly test one. One of the big struggles we see is the designing of the program: what earning rules to have, what rewards to offer. Brands can take months on this. That's why we built this AI loyalty designer. The main benefit is that you don't have to think about the earning rules, reward types, or how many discounts to give. The AI builder designs the entire program for you, and you can just launch it and save time.

Matt: That could be helpful to many brands. My experience with brands designing loyalty programs is that they look at what their competition is doing and copy it, maybe with slightly better discounts or more points. Would you say that's a good idea?

Marton: I wouldn't. It's the same problem as transactional points. If you become the same as everyone else, nothing makes your customers want to stay brand loyal. One other thing: if you don't want to use the AI loyalty designer, we provide complete loyalty program recommendations to the brands we work with. We usually push for an initial call where we discuss the type of products you sell, then come up with an idea, and we can even set it up for you.

Matt: So it's AI plus a human in the loop.

Marton: Exactly.

Back to Sidekick

Matt: I promised I'd get back to the Sidekick thread we opened at the beginning. I was wondering whether Sidekick can do this job, designing a loyalty program based on what it knows about the store. The results were interesting. My observation is that for this to be effective, Sidekick needs to know not just the prices of the products but the COGS, the cost of goods. Then it can calculate gross margins effectively and make suggestions on how to structure the program. So connecting Love Loyalty to Sidekick will help a lot of brands design programs in an effective and profitable way.

Marton: Exactly. Sidekick is getting more and more clever, and I think it's a must for modern loyalty programs to be integrated with it, because in loyalty it's so important to understand what's behind it.

Matt: What I see with running e-commerce is that many people treat a loyalty program as a one-time task. They set it up, check it off the list, and move on. Maybe it's a matter of awareness. Some brands pay a lot of attention and experiment. Or maybe it's that in Europe, customer acquisition costs aren't as high as in the US, so people don't pay as much attention to retention. I suppose that will change soon, and having the right tools will give you better data. It should be easier to ask Sidekick, "Can you send me a report on my loyalty program every other week or month?" Eventually you'll probably have an autonomous agent that can adjust the program's settings for you, so you won't have to pay attention. But I think having a human in the loop would be important here to make the program effective, right?

Marton: Yes, exactly. We're getting to a stage where people will want to know more about their retention metrics and their loyalty program, and it'll be much easier with AI. Hopefully more brands will not just launch a program and know nothing about whether it performs, but will see that if they care about it, adjust it, and experiment, it actually affects their revenue in the long term.

Final thoughts

Matt: Exactly. Marton, do you have any final thoughts or words of wisdom about loyalty?

Marton: I think it's what I said last time as well: all brands can have a loyalty program, no matter what they're selling. The first and most important thing is to really know your customers, because with today's tools you can set up any kind of loyalty program. You don't have to think about giving points for dollars. You can set up any type of reward system because of modern tools. But for that you really have to know your customers' incentives. So understand your customers' needs.

Matt: Marton, thanks for this sequel. I suppose we'll see each other soon and elaborate on this topic again in the near future. We started with football, so I think it'd be fitting to close with it. Give me your podium, your top three.

Marton: Top three. Argentina first, I think they'll win. This is tricky because I'm not sure what the knockout ladder looks like, but if I had to say, Argentina wins. Second, the Netherlands. And third, if it's possible, Spain. What's your podium?

Matt: I think Argentina will score all their goals with Messi only. But I can imagine the same final as four years ago, France, and then Spain. So Argentina, France, Spain. Yours is probably that, or maybe England instead of Spain.

Marton: England, I forgot again.

Matt: Right. We'll see. Marton, thanks again, it was a pleasure talking to you.

Marton: Thank you for having me.

Matt: Thanks again. Bye.

‍