BizTech Forward: AI in Retail & What Shoppers Want

Anni Tabagua: Hi, everyone. Welcome back to BizTech Forward. I'm Anni, and in these next two episodes, we're talking about AI. I know it's really original of me, but this time I want to focus on something specific, something that actually has to do with real life. Not just theory, but the places where it either works or doesn't. So, I was thinking that retail may be the best testing ground for this. Also, because retail is very close to my heart. When I'm not working, I spend all my time on fashion resale platforms. So, yeah, we're told technology is transforming shopping. Everything is going digital. AI is personalizing our every click. However, we still go to the store. I just want to know what's happening and what's going to happen. To discuss all this, I'm joined by Dmitry Olerinskiy, Digital Strategy Director at Decathlon, a global sporting goods retailer. Dmitry has been at the crossroads of retail and tech for a long time, so he's actually a great person to talk to about all this. So, Dmitry, thank you so much for taking the time for this.

Dmitry Olerinskiy: Hello, thank you for having me here. Huge thanks for such a nice introduction.

Anni Tabagua: Yeah, yeah, it's a pleasure. I'm excited to talk to you. Before I jump in and ask you some serious questions, I want to know how many times you have heard the word AI today?

Dmitry Olerinskiy: You're the first one.

Anni Tabagua: Oh, wow.

Dmitry Olerinskiy: It's like, yeah, it feels like with every hype, at some point, people start trying to avoid certain words, just not to buzz them too often. So, yeah, not for today. I'm trying to avoid it as well.

Anni Tabagua: Okay, so in the next maybe 20 minutes, it will be impossible to avoid it. Sorry.

Dmitry Olerinskiy: Well, we'll utilize my quota for AI words for the next week, but that's okay.

Anni Tabagua: Yeah. So, I’d like to start by asking you some general questions, but since you've been working in retail tech for a while now, I'd like to know what you’re seeing right now. Is there a gap that you're noticing, like, between what retail companies are doing with AI and what's actually working for customers? Or am I wrong to assume there is a gap? Is technology, or AI, delivering everything that it claims to deliver?

Dmitry Olerinskiy: It's a very interesting question. I think it relates much more to the adoption curve of any innovation. So, when we're talking about retail, especially customer experience in retail, it's very much about not only real life, but also that part of real life. And everything should work smoothly. And when everything already proved its efficiency and its validity, when we're talking about AI, we're talking about something super innovative, something emerging at this particular moment in time, meaning there are so many things which not only haven't proven themselves yet, but also just appear – people are still exploring their capabilities and their function, possible applications and functionalities. That said, it's sort of by definition, of course, certain that in the field, in the store, whatever that store is, for the moment, we don't see absolutely all capabilities of AI we are already aware of. And this is good. On the other hand, it would be much worse if we were to start deploying every functionality we find immediately. Between the moment we discover the functionality and the moment it starts working absolutely reliably, there is a certain test and a specific timeframe. And if we start showing the public everything we find, people get much more confused and upset rather than happy. Because not all experiments work, it is not surprising that not all are successful. However, as a matter of fact, I believe that we are already starting to see numerous real-world applications of AI, to name just a few. There are a lot of applications of AI in customer service. There are numerous applications of AI in helping consumers with their selection, including sizing and styling, across the board. And these are all real-life cases.

Anni Tabagua: You mentioned this word "timeframe." And I wanted to ask you, what should be a realistic timeframe, because I feel like when we're talking about AI, nobody has patience. Everybody wants everything done now and done, you know, really well. So, what is the realistic timeframe?

Dmitry Olerinskiy: Depending on what exactly you are talking about? It's like for every technology or for every application of technology, there is a different timeline. I believe that at the concept level, it just makes sense to give ourselves enough time for experimentation, so we don't deploy what's not ready yet. We don't eat uncooked bread or pasta. We let it cook properly and then serve it at the table. Therefore, the same approach should be applied to AI as well, to any technology. Basically, I believe for both of us as insiders working on the technological side of things, it's very natural to meet technologies much, much quicker than regular consumers of a Decathlon store or any other store, simply because we've chosen our profession to face this novelty first and foremost, and we've chosen our profession to help those technologies land in the best possible way to customer experiences. So, we are those who are expected to give it a try and conduct all possible experimentation before we finally shape it into something that deserves to be shown to consumers.

Anni Tabagua: Yeah.

Dmitry Olerinskiy: Which takes time. Which takes time. And I mean, sometimes it's only a matter of refining the approach, and sometimes it's also a matter of technological development. So, for example, with conversational commerce and with all those things that just start appearing, they were unimaginable very recently. They just appeared, and there is no question that it will take time before all retailers adapt their technologies and all the connectivity protocols are built and set up. Therefore, these technologies will begin to be actively used by customers.

Anni Tabagua: Yeah. I don't know why I'm stressing this so much, but I just really want to know, like, time, and also, you know, are we talking decades or are we talking...

Dmitry Olerinskiy: The funniest thing about it? I recently heard, or literally just today, in one of my other conversations, that the guy was pulling a simple calculator from his pocket. I don't know why he carries a calculator in his pocket. I didn't manage to ask. I was so astonished. But he was like, you know what this is? Also, more than that. This is a generative AI because when you're writing something like 4x4 equals something, you're essentially creating a prompt. And I thought, ‘Jesus Christ, I never considered this from this perspective.’ Right. So, I think that a big part of that hype also depends on the words we attach to that and the meaning we're trying to stuff into those words. In reality, we're talking about a very complicated scope and set of technologies, and we have to be – if we want to be seriously specific—we need to be very careful with what we name with the term AI because it was described many decades ago. It was used here and there, bits and pieces, for quite a while. And now we're entering – we're just, if you remember this innovation adoption curve, in terms of general knowledge of AI and in terms of general accessibility towards AI, we're just climbing up that slope. And since more and more people are becoming familiar with the term AI, and as more and more people start having their everyday experiences with something they call AI, it becomes increasingly inflated as a hype. However, from a technological perspective, it's a completely different matter. And getting back to your question about the timeframe, it's not that scary in real life, as soon as you start considering very specific applications.

Anni Tabagua: Yeah. That's encouraging. However, I keep thinking that we've had e-commerce for so long. We offer AI recommendations and virtual try-on features. There is an app that can tell you what color suits you. You know, you name it. However, the traditional store model remains effective. People still go to the store a lot. Yeah, I want to know. Yes. And I'd really like to pick your brain on this and give you the space and time to walk us through it. What do you think about it? Why do people go to stores? Why do they need it? Why does the traditional model work? And where is this going? I'll stop with the questions for now. But as a side note, I read a lot of Substack newsletters about fashion technology, you know, retail and all, just because I'm interested and I see all these gift guides coming out lately, you know, for gift season and a lot of influencers, let's call them that, and fashion writers encourage people to go to stores and, you know, they encourage like right now like never before that, hey, you guys, please go to the store in real life and ostensibly like nothing beats in-person shopping. However, despite everything we see technology doing, I would like to hear from you what you think. Why is this happening?

Dmitry Olerinskiy: Because people are social animals. I think that there has been extensive research done on that topic, and numerous observations have been accumulated over the years. We're very social animals. We need interactions. We need other people around us. Additionally, there are various types of fun, entertainment, and leisure activities, and shopping has become an increasingly popular form of leisure. So, luckily or unluckily, it got so far from the transactional meaning of it all the way into just one of the forms of entertainment. So, I don't expect people to stop going shopping anytime soon. Why would they?

Anni Tabagua: I mean, it does make sense what you're saying. It's common sense. People need people. People, you know, need communities. Yes. However, I still feel like I need something else to be truly convinced. Because when my dream shoes are one click away, right? Why bother? Like, why? Why do I go through all this shopping?

Dmitry Olerinskiy: There's the whole plethora of reasons. I think that this is a very interesting topic to observe. When we say to ourselves that my shoes are one click away, I believe that everyone can notice two completely different scenarios. And one scenario, you're just making that click, and you're getting your shoes delivered in a couple of days. And this is just a done deal. It's a pure transaction, and you're happy with your shoes. In the other scenario, something entirely different happens. You don't make this click. Rather, you keep processing that thought about a shoe. And maybe you'd like to see another shoe. Perhaps you would like to look for not just shoes, but shoes and bags, or perhaps something else. And you end up basically entertaining yourself with this choice. And I think that we have to keep in mind that choice sometimes is not a purposeful act, it is just a form of entertainment by itself. And that's totally okay. I mean, that's sad. That's, in a way, pathetic that we need to create an artificial need for choice to entertain ourselves. But, practically speaking, I think it's important to remember that the choice itself is a very interesting entertainment.

Anni Tabagua: That's interesting. Like, I cannot argue with you on this, because if you frame shopping as entertainment, then, yeah, people want to go and get entertainment. However, it also occurs to me that, if I view it not as entertainment, but as a means to an end, where I know what I want and I want the most convenient way to obtain it, then I wonder why people go to stores. But if you insist that it is a form of entertainment, then I'm like, my hands are tied. I just hope that...

Dmitry Olerinskiy: The Nobel Prize that was given in 2016 for a statement that there is no homo economicus. I mean, I'm oversimplifying things, but even economists who believed for many decades and centuries that people are very rational finally agreed that people are far from being rational to an extreme. So, yeah, we're very much more emotional. And of course, we're going shopping, not just to get the stuff we need bought and delivered at home. We're also getting a lot of emotional support. And I mean, getting back to the initial subject of the conversation today, I believe that this creates one of the strongest amplifiers, and at the same time, one of the highest possible hurdles in the integration of all possible AI applications into the shopping experience. The simplest way to consider AI in terms of the shopping experience is to view it as very functional. It's like simplifying the practicalities of choice and selection, incredibly straightforward and very easy, done long before conversational and generative AI. Done before with simple filtering. Not commonly used in real life, but widely used in the online and digital world. Very much used in the store. Nobody uses screens whatsoever. Everybody likes to talk to a person or just browse through, which means people choose not only by filtering information. People also choose perceptually. And the key question, even when considering fitting and styling, is what? Exactly those moments and exactly those places on the retail floor. And what shall be the shape of these AI tools so that people will pay attention to them and spend enough time with them? So, a space will be created for advice. That's a significant question that many companies are trying to solve. And the best possible approximation we've got so far is those AI-backed kiosks, like digital kiosks which start talking to you and you struggle to understand what you want this piece of AI to do for you. But I believe that the underlying challenge there in this specific case is exactly the fact that people are by far not only rational, but also emotional and need much more than just factoring out the important criteria.

Anni Tabagua: So, I guess it's even wrong to ask, hey, is AI going to replace, you know, physical stores anytime soon? Because that's not a goal at all. Right?

Dmitry Olerinskiy: I don't think so. Yeah, I don't think so. I think the internet will never replace TV. Or like TV never replaced printed media. They somehow manage to coexist, right? I mean, I'm not saying which channel gets more advertising revenue. That's out of the scope of our conversation. I'm saying that all those media channels coexist, and I believe that all purchasing channels will also find their ways to coexist. Exactly because they are much more than just purchasing channels, they deliver a much broader scope of value than just parcels and goods being available for shopping.

Anni Tabagua: And also, when I think about the shopper of the future, I was wondering what AI should definitely be helping shoppers with. Like, I'll give you an example. If I'm buying running shoes at a store, should the store also provide me with running tips and route recommendations through AI? Because I was thinking that maybe I would trust the suggestions coming from a sports store more than from ChatGPT, because they at least know what I'm running in. So, I don't know. What do you think about that? Like, the question is, what should AI be helping with?

Dmitry Olerinskiy: Keeping customers' hope? I think, how can I say it? I think this question is rather interesting by itself because it tends to pack too many things into the same basket when we're thinking about running shoes. In your example, that's a very strong inference and a very long shot to say that customers are also interested in route advice. In real life, the majority of people who buy running shoes are either walking in them or using them for casual wear, or they very much know where they're running. They don't need the advice; they just need a shoe, right? So, I think that in simple terms and simple cases, it's much more practical and trivial. For example, running shoes are a good example because we all have slight differences in the way we move our legs while walking or running. And with the longer distances you run, the more—not only these strains—but also the more important it is to know what exact type of, I don't know, a specific medical term, what exact shape of foot you have, which type of shoes you need to wear. This fitting can be highly personalized for extreme professional athletes. But for regular people, this fitting also provides a lot of value. Many brands produce a variety of different types of shoes, focusing not only on colors and models, but also on the specific requirements of the human body. And this automation of selection may be done probably faster. At the moment, it is done by recording on the video camera—you run on the treadmill, somebody records you from the back, from the front, and from the side, and they browse at low speed your videos where it's visible how your legs move, and they give you advice on the shoes. This can be automated with AI, and you don't need generative AI for that. You need a very clear algorithmic AI. However, what you need from the AI side is the capability to process large amounts of data and interpret video recordings in real-time. And this is a very realistic case, which holds a lot of value for regular people, but it's clearly not as elaborate as something we connect with holistically, such as experiencing the purchase of a shoe and your overall running journey over the year.

Anni Tabagua: Yeah.

Dmitry Olerinskiy: We will develop by making small steps. I think we will develop the most vivid and natural introductions of technology into our everyday life, and then gradually build on that.

Anni Tabagua: And is there a lot of talk about making things even more personalized for customers via AI? Because often, it feels like algorithms are just guessing. Some other time, it feels a bit creepy and intrusive. And I wonder where that line is between, you know, what is useful and this is a bit intrusive, or what is helpful versus what is not. AI is failing that.

Dmitry Olerinskiy: I think in this conversation, AI is just another tool. And the real balance happens between two forces. On one hand, we have a monetization approach – companies need to sell more to deliver on their targets and promises to investors. On the other hand, we have customers' tolerance towards intrusion into their personal lives, as well as their tolerance towards all possible requests regarding their personalized data that can be processed, and customers’ needs for personalized goods. And those two forces balance each other. There will always be a push to make advertising more personalized because personalization from a business perspective means higher conversion rates. However, there will always be a pushback, whether passive or negative, from a customer’s perspective. They will simply vote with their wallets. They will stop buying. If it becomes too annoying, they will stop buying. If it becomes too intrusive, they will just balance each other. With or without AI, they will continue to balance each other.

Anni Tabagua: And if we talk a little bit about implementation, I wonder if somebody, you know, or a retail company is listening to us, and they are having these conversations internally that go like, hey, we need to do something with AI now. Like, what does that mean? What usually happens, or what is the first conversation that they should have, but perhaps they don't usually have? In other words, I want to hear if you could comment on some common mistakes that you've noticed and would advise against.

Dmitry Olerinskiy: I really like the first wording of the question. The first thing to think of is the complex processes that seem to be automated. The second thing is exactly where you need to process large amounts of data on the fly, or generate all possible options of answers or responses on the fly. This is where generative AI may come in handy. I would always start from the business need. I will always start from where you are in your customer journey. For instance, in the retail context, you often see black holes where customers are lost. And it may be something very simple. It may be something as simple as in-store navigation, because for larger stores, navigation inside the store may become a challenge at some point. And we have in-store navigation signs on purpose. And if you discover that there are some parts of the store that are rarely visited, one approach may be to create a large poster, but another approach may be to introduce some form of digital navigation. If you have an application, introduce navigation across the store in the application and see how it works.

Anni Tabagua: Yeah.

Dmitry Olerinskiy: So, I would always start with the challenge, start with the pain. And better if it is the customers' pain.

Anni Tabagua: And then think about the foundations.

Dmitry Olerinskiy: And then it's not necessarily that AI will be the right answer to the pain you are facing. Start exploring the pain and start finding the best possible solution. Knowing that AI, since generative AI appeared, I mean, since it appeared, it just enables the whole new set of tools and capabilities. But they will not replace the clarity in terms of what exactly the pain is. It will still start by understanding what exactly the customers are willing to have, and what they currently lack. And how can we ensure that customers start getting what they're looking for in the smoothest and simplest possible way? And then maybe in some cases, you will discover that AI can be a solution.

Anni Tabagua: Right. I also want to hear your thoughts on the human side, not just the technical side, but when you observe other companies and notice some mistakes, what is happening? Yeah. Like, you know, what is an organizational mistake? Or, you know, more about the human and all this discourse around AI replacing humans and all that stuff.

Dmitry Olerinskiy: There is massive fear that AI will replace whole professions and whole teams of people. I don't believe it that much. I see that AI currently serves as a significant complement to qualified professionals, to put it that way. I see a lot of attempts to implement AI for the sake of implementing AI because this is the hype, and I think that this is one of the simplest ways to make the best possible mistakes, because usually, a case should come first, and a decent business case and user case should come first. And then the rest is boring. It's as if change management is a project in itself. Change management is a whole work by itself. Implementing AI, especially given its consequences and repercussions on overall business processes, typically requires significant change management, which should not be overlooked. But this is rather the classics of implementation of any new technology, right?

Anni Tabagua: Yeah, this is really interesting. I like how direct and candid you are with your answers, and we're slowly getting to my favorite part of the conversation that I like to wrap with, and it's more of, you know, the talk about the future and what is ahead. But before I ask that, I want to know if you have any unpopular opinions on the topic, something related to AI or retail shopping that you think is a hot topic you could share.

Dmitry Olerinskiy: On AI and retail?

Anni Tabagua: Yeah.

Dmitry Olerinskiy: Well, yeah, as I said, retail stores will not disappear, and people will keep shopping offline as much as they do shop online.

Anni Tabagua: Do you think it's an unpopular opinion? Because when I asked you that question, you seemed surprised, as if to say, “No way.” What do you mean?

Dmitry Olerinskiy: I don't think that this is an unpopular opinion. No. It's hard for me to take on an unpopular opinion because I usually don't pay that much attention to the balance between what I'm thinking and what the real popular opinion on the subject is.

Anni Tabagua: I see.

Dmitry Olerinskiy: I don't care that much. But in terms of AI, I do believe that we will see a lot of implementations and a lot of experimentation around the topic going forward. I don't believe that AI will replace humans to a degree that we should be really scared. I believe that AI serves as a—and again, I mean, AI is a very broad thing, generative AI and the thing which more and more people start being acquainted with—I believe it serves as a huge booster to our awareness about what actually we are processing and which vocations we're choosing and pursuing in our lives. And I feel like previously we have been challenged by the number of professionals in the field, but all those professionals were humans. Now we'll start to see that AI is not as professional as our peers used to be. AI is something that can be taught and educated to become anybody. And I feel like we got a portion of a totally new type of motivation with AI appearing, and we are sort of invited to keep developing ourselves and to keep being creative, keep being hungry and curious, and all those types of things for the sake of not only keeping professions alive, but for the sake of not missing real opportunities. From a progress perspective, AI being capable of taking over some mundane tasks from humans does not represent a threat, but rather creates an opportunity for those humans to free up some of their time. And the real question is not about the quantity of things that AI will overtake. The real question to me is if people are capable of doing something real with the time that all of a sudden, they will have freed up by AI. It's just an invitation to honestly answer what's interesting enough to pursue next.

Anni Tabagua: And if you think about next year specifically. So, who knows, maybe at the end of next year I'll reach out to you and say, ‘Dmitry, remember what you said a year ago?' So, I'm setting it up right now for next year. So, what do you think is the most realistic thing that will happen next year, you know, with AI and retail? And it might actually change how people shop specifically for next year, maybe.

Dmitry Olerinskiy: A very interesting dynamic is going on with this AI-driven agentic commerce, conversational commerce. I feel like this will either boost or disappear. Another interesting dynamic is driven by the take rate of marketplaces and the AI challenge they present to sellers. This dynamic may also lead to standalone brands rising again, which will, in turn, reshape the digital retail industry landscape. I think that one year is too short a horizon to see any significant changes, unless there's some sort of revolution or serious crisis. However, what OpenAI has achieved with conversational commerce, and what is currently ongoing with the decrease in costs of all those deployed model deployments, appears to be a significant boost to technological development and improvement of the technologies applied in retail. I hope to see many interesting cases, customer-facing, that provide real value in terms of a positive user experience on the floor.

Anni Tabagua: Well, great. Let's touch base again next year.

Dmitry Olerinskiy: Next year we'll check. Yeah.

Anni Tabagua: Yeah. I'll remind you. And last thing, you personally, what makes you genuinely excited about working in the space right now?

Dmitry Olerinskiy: It has a lot to do with people.

Anni Tabagua: Please, please tell us more.

Dmitry Olerinskiy: Of course. I mean, retail is a very interesting industry because on the one hand, it's very commercially driven. It's all about merchandising, and it's all about making money. At the same time, it's very detail-oriented. They say retail is detail, which is very much so because there are so many things happening at the same time. But the underlying truth is that retail exists only as long as people are there and shoppers are out there. And at the end of the day, we are there as long as we can deliver on customers' expectations and continue to excite our customers. And I find it very creative and challenging at the same time. Yes, you are there to develop techniques and improve the customer experience you are having. But at the end of the day, customers are either happy or not. If customers are satisfied, they will return. If customers are unhappy, they are unlikely to return. And it's as simple as that. And the existence of people at such a short distance towards whatever you're doing, I find very fascinating. It feels very real. It feels very purpose-driven in the best possible way.

Anni Tabagua: Yeah. So retail is detail, and retail is also people, right?

Dmitry Olerinskiy: Of course. People serving other people.

Anni Tabagua: Dmitry, thank you so much. This is really exactly the kind of conversation I wanted to have. I really, really enjoyed talking to you.

Dmitry Olerinskiy: Thank you very much. Thanks for having me.

Anni Tabagua: And to our listeners, thank you for tuning in. If you have any thoughts, insights, requests, or questions, please don't hesitate to reach out to us at biztechforward@dataart.com. That's it for today, and we'll see you next time.