#4 MAARTEN DE NEVE & ANTHONY IGLESIAS GUERRA - Unpacking hyper-personalisation in marketing

(upbeat music) Hello and welcome to the Bam Deep Dives, a podcast series by the Belgian Association of Marketing. I'm Sarah from the MacKina, an E-Pam company, and I invite experts to come and have a chat about some very hot topics in marketing. In each episode, I make them dig deep, so they give us practical insights and real world case studies. Today, I'll be talking with Anthony Iglesias from Kervur and Martin the Navy from Immaculam. What we'll be talking about, hyper-personalization. I hope you're ready, let's get started. (upbeat music) Hello Anthony, hello Martin, welcome. Could you quickly introduce yourselves and you can start Anthony? - Thank you Sarah. So my name is Anthony Iglesias. I'm working at Kervur since one year and a half as a personalized customer marketing director. So what's in your name? Long story short of our span of responsibility cover, the personalized direct to consumer marketing. So the direct relation with our customers, but also our loyalty program. So we try to offer the best loyalty program we can and we animate it early and long. - Interesting, and you Martin? - Thank you Sarah. So my name is Martin Teneva. I'm digital marketing lead at E-MacKina. I've been guiding clients and consulting clients in direct marketing or personalized marketing, whatever name we give it in the past 20 years. So I'm really passionate about one-on-one marketing and making sure that we create this experience that create business and value for our clients. - Interesting, well, I know that the two of you are here to talk about hyper-personalization. So Anthony, can you start the conversation by giving us a definition of what hyper-personalization is? - Yeah, sure. So to me, hyper-personalization is the level of personalization that is created when using a combination of advanced tools, concepts like AI, real-time, whatever you have in mind. And some multiple type of data. Sociodemo, basic behavioral profile, declared, observed, calculated. So when we combine everything together, then you get the closest to the hyper-personalization. - Martin, do you agree? - Absolutely, I think it's a perfect intro to say that personalization or hyper-personalization is the advanced form of personalization, wherein the past where we were doing personalization as well. But more in a project approach, not automated, using less sources of data. And these days, we have a lot more data and technology at hand, so we can improve and go to that hyper, even if it's a buzzword or one-on-one adaptive way of using personalization. - Yeah. - Is it really clear for everyone what the difference is between personalization and hyper-personalization? - Well, that's a good question. I think the main difference or the main uplift from personalization is it is automated. It is real time. It has all sources of data that you can use and it's implemented in the full customer journey. So not just in the marketing journey, or but it's the full journey, including even sales processes, custom service processes. So it's the whole user experience that's been influenced by personalization. So it's an ecosystem? - It's an ecosystem and it's also a level of personalization. So you can have different level, not in a particular order, but segment-based, behavioral-based, predictive, location-based, whatever. And I think when you can use wisely all those different type of personalization, you can get to the hyper-personalization. Of course, I said wisely. If you go to foreign, you don't have to look creepy. (laughs) So what do you put in place to make sure that you don't become creepy as a company? - Well, at KFOR, we send some more less 300 emails per year to our database. Not everybody received 300 emails, of course. We try to avoid spamming. - Let's hope not. - But we send 300 emails. So some become day-to-day job for us. In those ones, we don't need to go that much in-depth. Imagine you're working in the street, you meet someone you don't know from anything and he starts knowing everything about you. You would say it's a bit creepy. Of course, on the contrary, you meet a friend you haven't seen in a long time and that person starts asking you question and start interacting with you, then you get in a comfortable situation. That's what we try to find the right level of personalization. So, as I said, not every email has to go to find a personalization. It's not an objective as such, but it's important to find the right level of it and of course, plan it. The more you plan, the less effort you make for every interaction you have with your customer and the game is on. - I hear that hyper-personalization is not the end goal. - Indeed, that's one of you, of course. On the one hand, it takes a lot of effort to go there. You cannot automate everything and, as I said, when you start the conversation with a new member of the program, even if you know more information than you should know, I think it's nice and wise to use the level of personalization that the customer feels comfortable with, they're more than more so, they're mister, these are that, can be a good start to talk with. - I want to bounce on that because you said the level of personalization that people feel comfortable with, how do you identify that? - Experience, I would say. Try and test, and then you start, like, actually, I always try to put myself in the shoes of someone I meet, like the example I give in the street in a bar or whatever. You don't start immediately with, I know you're on 20 years old, you leave there and you start with, hello, my name is, these are that, who are you, and you start into an conversation. And that's, I think, the secrets of finding the right level. It's indeed, it's not an end state to get the fullest out of that technology or out of that tactic. It's not a purpose to go to the fastest possible in high-purpose personalization. It's trying to make your user and your brand experience the best. So it's like, the moment that you feel that it's no longer advertising, but that it's helping you and that it's relevant, then there's probably no problem in sending one communication more than yesterday. It's the moment that you feel that, okay, they're using these tactics to go all the way and it's beyond what I was expecting. And they weren't really transparent in the data that they needed to do this. In that case, you feel like it's becoming creepy. So it's the same as in advertising, advertising interests you. If it's interest you, then it's relevant. If it's not interesting and if it's really getting creepy, then you would say, okay, this is the border for me as a person, we cannot express that. - We've now talked about the negative, which is the creepy part. But when do you know that personalization is actually successful? - Well, for me, personalization is successful when you really feel the value that it brings. And it's not just, you have a brand experience, you have brands that need to make sure that the user experience is delightful. The moment that we can say, this is really now bringing me extra value, then personalization is good. Imagine that you work in a sports brand. The moment that you feel that this sports brand is really guiding me as a person and bringing me relevant information that helps me uplift my performance or my user journey. That moment, it brings value and you feel okay. This brand really knows me, understands me and communicates or reacts upon that. I think that's the goal to really bring that relevant value. - And it's careful. - We're lucky to be in a business where we can measure a lot of things. And of course, most of our interaction with customers goes with a coupon. So it's easy to measure how successful a campaign will be. We know the cost, we know the habits of the customer. We know if it puts additional turnover or not. So I think that definitely a couponing would be a good start to measure if personalization worked or not. And we can always compare with people that didn't receive that personalization and how the sales increase. Because of course, we don't always only do personalization. We also have other activities and sales increase in some situations, hopefully the most of the time. But you can, we have tools to measure how good one specific action will be. - How do you personalize coupon? - Personalize coupons is based mostly on the behavior. So on the product based, if you buy or not specific products and sometimes also on the total basket. So we send a mailing every month, for instance, a direct mail paper mail, where we send to that audience, we call that a total Acha, a coupon based on your total basket, where we try to ask you to make one additional effort. Of course, we are very generous, we're going with it. We try to create an habit of having bigger baskets. And on top, we also offer four different coupons that are more product-based. So if you buy a specific product, you will probably get the coupon that's to thank you to buy the product. So it's not always and only triggered to make you buy more. It's also to thank you to be a good customer. - I think it's important indeed to thank you on what you did in the past. Maybe see, okay, this is a product that interests this person. How can we make sure that this person can buy more or buy more with a promotion? But you also need to think about how can we inspire that person, maybe that person would also be interested in a next best product that is really linked to the previous product. And secondly, certainly, maybe there's also the possibility to predict when a person has a specific frequency in buying products, we can easily say it might be that the next time that he buys is in that period. So we can do a suggestion there. And at that moment, you show that you're relevant as a brand and you're showing that, okay, we understand you. There are specific needs you have that come back, but there is also inspiration that you're seeking for. We bring you the inspiration that you need in the communication that we sent out. - And indeed, if I can add on that, I always take the example of a airlines company. The more you fly, the more you get miles, the more you fly again. Versus the telco company, the less your customer, the more they try to attract you with a lot of rewards. We try to be a bit in between and of course, you can move the hour a bit left or a bit right and also make a difference between customers. Not all customers are the same. So these are more rewards. Some get the opportunity to buy products that don't know because they buy the competition. We don't want to make people buy and eat more than what they regularly eat. We try to get a market share from competitors as well. So we try to play with all those different aspects. So as I said, we have those regular mailings, emailings every month, every week. But you also have customer lifecycle program. So for instance, we'll come and teach on birthday. Just name it. Where we try to adapt our message on one hand because it's not only about couponing and buying. It's also about communication. So we try to adapt our message or call to action like coupons and the rewards all year long as well. Because it's important to be consistent in what we offer to the customer. I hear a lot of variety in the way you address the customer, right? So is that also part of hyper-personalization? What do you exactly mean with variety? Well, for example, we send you coupons based on your behavior, the past purchases. But we also send you an email on your birthday, for example. Could be indeed. And I think that's what's passionate me. That's why I'm passionate about hyper-personalization. Because in the end, it's a really interaction with every single customer. I don't think that's too customer. If you compare what one customer receives and one another customer receives a nearly based, I don't think we have two customers that receive the same sets of information, of communication type, of our rewards, everything is different and we try to adapt ourself to every different customer. It's not only another person, it's also contextual. So a person can also change. It can be the same person, but in another context. And maybe interacting with other channels then and other moments. So you need to think about how is this person in what situation is in. How can we be relevant in that context? So it's even within one person that you need to change your approach of them. But this kind of personalization really hinges on the data you have. Absolutely. That's key. I think if you, that's the main thing you need to think about, if you have a data foundation, at least something to start with, then you can go and approach personalization. And from that data foundation, you can then further build on a nurture every data point that you might need in the future. But that's what we often see with clients. So there is a foundation that is already there. But it's maybe a bit siloed. It's scattered over different teams. People have different understandings of the data points. So that's mostly what is needed at first to make sure how can we approach data in a unified way and then start from there on. You don't need to wait until you have that roll throw as of data. You can start already with what you have today. Yeah, totally. When you set different understanding of data, I remember when I started at Carfour, I discussed with my colleagues and every person I talked to, I asked, who is the customer of Carfour? I think I talked to 10 person and I got 10 different definition. You talk to e-commerce, you talk to branding, you talk to anyone to the stores. We have three type of stores, hypermarkets, supermarkets, and proximity stores. You talk to anyone. You have a different understanding of who our customers are. But in the end, it's just in person with different shopping missions, a different moment in life. Some have children, some are getting children, some are parts that are played. So it's really complex, detailed but at the same time, very interesting and I love it. I think it's not so uncommon that teams within a company don't have the same definition of a profile. That's mostly what we as an external company try to do, is to make sure that we have one unified definition because we've seen even the most mature companies who would have really advanced personalization or market automation, whatever you name it, the different teams at different approaches. For example, there was a web personalization team and then an email, so they were split into silo channels and then you say, OK, these people don't interact with each other. How can we make sure that there is one team, one approach and one definition of data? And that's, I think, really key for mostly the bigger companies to make sure that they get the most out of personalization. And then the logical question, me as a listener, my logical question is if you face this challenge, that the data is everywhere, or people don't have the same definition of data, how do you then bring all of this together? Well, what to be mostly do, it's not, of course, easily said because we have an outside view. The people inside the teams, they have specific team goals, et cetera. From an outside point of view, what we try to do is create some kind of agents, people who are responsible for data points. And so together with them, we make sure that there is a unified understanding of data. You create a view on the data architecture. You create a view on the data fields through a data dictionary. And then you implement use cases together. And just one simple use case to start from with the delivery in the next month. And that's the way we set our first steps into having a unified view on data, unified success. And from there on, they can easily build further on. That's how we approached it. I don't know if that's similar at the case. Yeah, totally. Before I arrived, I think it was three, four years ago, the decision was taken because indeed there was data a bit everywhere. The decision was taken to have a common data set. With common definition, the investment was done to have everything cloud-based, accessible to some people that would need different type of data. We invested also in data governance to have one clear definition. So we have a team of people that work all day long, all week long, to just make sure that the definition is the same, reporting or consistent. It's not always easy because anything can be looked from different perspectives. And it's natural. We just try to be the closest to each other, or at least an explanation why we look differently. That's the same data. The challenge for me is in a company such as Carfour with such huge data sets, where a project is not a small project that is in a couple of days, trying to follow the technology that goes faster and faster. So there, I think, there is a challenge for every big company and every big database that needs to evolve continually. What is the biggest challenge internally to get people to your night? I would say first is already get rid of the silos. Everybody has its own objectives. That's an all the different objective answer to the same company objective. But still, everybody wants to go fast, wants to be the best in what he does, and doesn't always take a step back to look at the impact of what everyone action has. So I think that would be the first step to not only align objective, but align on the how. And the how is very often forgotten, I think. And then I would say, choose a process. So there are people that are there to make sure that the definitions are the same in our shared. So these were made to make sure that we all agreed on those rules. We have to agree that we don't change rules every two days. And seeking to the rules is sometimes important. Of course, we keep in mind that agility is very important because the world evolves, the customer evolves, the business evolves. And it's all about balance. Again, finding the right balance between what we can do, what we can change versus what was decided to stay there for longer. And I think for that reason, it's important to have, you need to have a long-term vision, of course, a roadmap, but it's really important to have a short-term approach to it. And that's what we do as, so each team, it's naturally, each one has his own objective. You do paid media, you do owned media. You have campaign people who work on one specific campaign. What we try to do is make sure that you have use cases that integrate all of these. And that they understand each other's objectives. If you have Roas to achieve and you have customer lifetime value to achieve, how can we bring those together into one use case and build from their own in multi-experts agile teams? - Also, if I, we've covered the definition of hyper-personalization. And then we started talking about the foundations. And we've now talked about data. But Anthony, you already mentioned it. Another aspect to hyper-personalization is the tech. - Yes, indeed. - So the tech, that goes a little bit beyond my direct scope. In the sense that in a company such as Carfour, we need to cut a bit more. Still, I have a seat at the table and we discuss tech together. I think that where every company is looking now, should look in the near future, is AI, of course. I don't think that many companies are there yet. We aren't there yet. We're still looking to how to use it, how to use it wisely. And what's different type of project we can use it for, as you said, use cases. So we're a bit in that situation right now where we look how to use it in the best possible way. - And I think most companies are in that situation. But the thing is, you see a lot of seminars of big technology players. And they have a hyper-personalization that comes from there. And they have this nice new technology that can do everything. And it's really promising. But the thing is that when you go back to the reality, it's often we need to work with your own current tech stack and build use cases from there. And from there, see where does the business value come from in the future? And do we need to adapt our tech stack? And if yes, how can we then build up to that? How can we build use cases that convince people to change the more tech stack? Because it's often not needed for each company at the level of maturity that they're in to change the more tech stack. It's always a belief that we need to do that first and then get value out of it. I think you can surely get value out of your tech stack today. It might not be hyper yet, personalization, but you can really build good value use cases already today. You've talked about the levels of maturity in personalization. I'm wondering how, if I had a company, how would I know what level I'm at? What are the levels and how do I know where I am? It's of course important to align with other people in the sector. So therefore, I think things like these podcasts are perfect. And you often also see different types of testing methodologies. I mean, all the tech companies have these maturity tests or some agencies have these maturity tests. You can do those. But what I mostly think is don't look at what the tech funders are promising you that will be there tomorrow. You need to look at your own stage of maturity and see how can I make the business value come out of that tomorrow. I think that's the most important thing. Because if you look at the long term, you will be mainly visionary and you need to act today. I think that's really key. Regardless of the maturity state that you're in, but I presume that every company has a bit of a view on where they are and aligns with stakeholders or aligns with agencies to know what the next step would be. And again, it says to make sense. Not every business needs the same level of maturity. So in the end, where you are versus what's-- so I think the most important is that you feel comfortable as a company to enter a discussion with your customer in the channel that you prefer or in the channels that you prefer, of course. But I don't see the level of maturity as a thing on its own. It's the capability that you need to be able to communicate with your customers and see that you have a positive return. Yes. Makes sense. We've sort of skirted past it. But I would like to circle back to AI. Martin, AI, you thought? Yes. Well, I think a lot of companies were already using things like machine learning in the past, or doing manual database work to make sure that they have the good segments of the sleeping clients, of the most promising clients of the loyal clients. We had machine learning to do clustering, for example, so that already existed. Now, AI is given a promise that it's all automated and in real time and it's fantastic. And it is. But I do think that we were already doing it and we shouldn't do it because of the technology's sake. It's not because of the fact that AI arrived that we should all be AI driven tomorrow. Yes, the future is there. But I do think that we need to look at the business use cases and make sure that see what we need tomorrow and see if the technology can allow us today to build it, knowing that AI exists. And if you need to do clustering or any other next best prediction, we can go there if it brings value. On top of that, a lot of the platforms already have AI driven out of the box tools that are available. Think of really, really simple tools like an email platform would have a subject line prediction or a decent time optimization. These are also AI driven and already really accessible for a few years now. Do they chase AI? First, I think you have to keep in mind that a tool remains a tool. A tool with the tool remains a full. And if I give you the keys of my AI car, you will probably not try faster than a pilot trained since years. This being said, I think that at Carfour, we are lucky to be part of a huge group. And we would be silly to not use that power we have. I think, as we say, AI is such a big thing, arriving, it's quite new. We have people that are at a level that can use to be able to negotiate things, take the time to think about how to use it. Of course, you want to have a seat at that table, to have our say. So we don't just let the group decide for us. We try to propose use cases, to influence, to tell them how much we could do with it also. But I think it would be a pity to not use the force of the group, Carfour, for that. Absolutely. And I see a very, very strong future in the agents models. So the agents will be there for us tomorrow to help us make our work, make us do our work better. And also to help clients, I think if we think about how can we use AI not only for efficiency, that's what we mostly are in doing today. How can we be more efficient? We should use this technology also to bring more quality. And then think about the last step, how can we use this new technology to build business models that really help clients? How can we build agents, for example, that help our clients to have a better customer or user experience? How can an agent help him going back to the sport brand? How can an agent help me to personalize my relationship with a brand and advise me on my next best step? How can an agent help me in being the best cook and finding the best matching recipes with the context I'm in today? These are the things that I think we should look on on a long term. You said agent, but for those of us who don't know what an agent is? Agents are AI-driven technology that allow you to do real-time automated services for you. Think of an agent. An agent can be also a multi-combination of multi-agents. Imagine that I want to translate my data into clusters. You can do that by an agent. Imagine that I want these clusters to be translated into persona that can be done by an agent. Imagine that I want to contact each persona in a personalized way on the relevant channel that can be done by agents. So the combination of multiple agents will allow you to create an agent-driven user experience. Talking about the future. But helping you to bring really contextual and relevant for user at the right moment, bringing the right message. And I think we should all think about what part of my job can tomorrow be done through agents? And how can we facilitate that? How can we do that in an automated and real-time way to make sure that our job becomes more efficient? And the user journey can then be more relevant. OK. I'm trying to picture it a little bit, right, this agent. So this agent is kind of a one touch point for you to then speak to other AI's or generative AI's or that sort of thing. So they, instead of speaking to a different AI for each thing, you combine them all through an agent. So I speak to one AI, and then that does several things for me. So first, look at the data, then see the trends in the data and then synthesize to persona, and then use those insights to give me content suggestions, for example, rather than having to do these full actions separately. I speak to one agent who then prompts the other agents or the other AI's to do that for me. Yeah. The agent that you have in mind is indeed a person, or a technology that you talk to, that guides you. What I meant with the agents is everything behind that makes sure that all the automated and real-time decision-making is done. You can split that up into different agents. And then indeed, in the end, you can have then, like an avatar or a personalized person that represents your company to translate that into servicing you. But I do think it's not about thinking about the channels that you're mostly on today, on which you communicate to a client. So it's like the classic advertising approach. We need to think about how can we make content in the future, delivered by these agents, in a contextual way, the right moment for a user, when he needs that information. How can we bring that content to that person at the right time? Interesting. We're not there yet. I'm learning. I think that if I take a step back on a more macro level, my gut feeling says that, in your first stage, I would use AI to get rid of repetitive work, I don't know, product description, the type of things. To free up time, to people to have more added value projects. Of course, if we look further, I'm sure I need to evolve by then. We would be able to get rid of complexity. And then I think that the third stage would be to go there. But I think we still have a lot of work to do and a lot to learn. It's a completely different approach and dimension. So I'm thrilled. It's an exciting future indeed. I want to come back to something you said earlier. And then you said the power that we have when we do hyper-personalization. And as we all know, with great power comes greater responsibility. So how do we hyper-personalize ethically? That's a subject we have every day on the table at Carfour. In case of doubt, we overkill. So we don't take any risk. We see some other actors in the market, competitors are not taking a lot of risks. And we chose not to go there. So all customers can continue to provide their data safely. We'll not misuse still. The temptation is big to push the boundaries. But I think every time we launch a communication, we try to put ourselves in the shoes of the customers. We try to use the data wisely. To not go too far in the personalization. To not use data in an ethical way like alcohol promotions to anyone. Also, baby, female hygiene. So there are some products where we are even more cautious. And for the rest of the data, we try to promote the coupons we try to promote. But it's a daily fight with DPO to use the right level of ethics and of legal compliance. I was wondering, and do you then have specific people who are dedicated to keeping an eye on that ethical matters? Well, when we need help, you can always go to legal to DPO. For daily matters, it's more on team level where we discuss, we have some experience as well. And I don't think one campaign goes off with only two eyes looking at it. So we try to help each other. Yeah, well, we see the same. So what we try to do is then create some kind of a policy model where each one has a responsibility or a task within a project team. And one of those is that really the person that represents ethics. So you always, you need to make sure that you always have one person who's dedicated to that for each use case. And so he or she needs to check the box and say, yes, I did a validation for ethical matters. Because indeed, the risks are too high these days. We have more data. We have a lot of data. So you should really always keep an eye on that. I'm happy to hear that you guys are doing that. I agree. As a customer, I'm very happy to hear that. I want to move to a little bit further into the future now because we talked, you know, hyper personalization. A hyper personalization is coming. We need data. We need tech. And we need an ethical approach to combine all of those to create value for the end user. If we step a little bit further into the future, what are the crazy things that we think we could do with hyper personalization? Can we dream for our businesses or our clients? I think you can dream very far. It's like what I just discussed earlier on. Mostly we recommend brands to start with use cases tomorrow, which you need to dream about the long term as well. And I do think a lot of the top brands, which you always refer to, you have the Netflix's and the Spotify's, etc. They do a lot of personalization. They have all the algorithms running to make sure that you have the right recommendation for you. They do think that this can all go a bit further because for now, this remains rather channel based. And they give recommendations. And there are two things that I missed there. And that's one context. And secondly, and I think you will like this is the serendipity. I really think that a specific context requires for me another recommendation. And if I always listen to genre X, a specific context would allow me to get another genre because I'm not in the mood of what I'm always listening to. That would be for Spotify. And secondly, we need to think about how can we make sure that we stay inspiring? And that's really a risk what I always see is that when you do recommendations, the risk that you're going to a bubble, a filter bubble, which doesn't allow you to really be inspired and go outside and discover new things, those, I think if we look at the future, I really would like to have a future where we have contextual moments together with brands where we get inspired. And it's not only recommending me the next best thing to buy, but also really inspiring me. And for me, that's the value that brands can bring. Well, if we allow ourselves to do, imagine that you sit in your coach, you wear glasses and you enter a shop personalized. So it's not even personalized communication as we do today. It's your complete customer experience as it's personalized. You see hours that you follow to get to your product, to your shopping list, but also to the product that you might have forgotten because we take that you wanted to do that recipe and you missed it on your list. We put some promotion forward. And actually, we don't talk anymore about high personalization in terms of communication as we do today, but we mix the entire customer experience. I think that would be a level that I would love to see one day, but let's see what do you guys bring us? Omni model experiences, I'm all for it. All right. A teeny tiny step back now because we dream very far. If companies have their data in order, they've organized the data, they've got a definition that is unified, they've gotten rid of their silos and they have the right tech in place. What else do they need in order to move forward towards a higher maturity level? I like the point of a two set about maturity level. So not everybody wants to go to the end, so and I'm convinced of that. I think what is mostly lacking is like the team approach. So the unified team approach, you talked about data and technology, but it's also about making sure that we have the maturity in the teams and that's more on how do we collaborate together to make sure that we not only look at our own expertise and our own channels and our own responsibility, but think in a unified way about the user experience. If you're saying indeed, I go to the shop and I want to be have the personalized experience there the same as I had on my app or on my website, I think that's the ultimate thing that you should go for. On any environment, any place that you visit or that you frequent the brand, all teams should collaborate and I'm not thinking only about the marketing channels or the store, but it's also about the sales in B2B, also about customer service. In an ideal way, you have a unified view on the data and the collaborative approach that each touchpoint blifts the experience. A collaboration model and also a specific set of skills and skills, I do think that everybody should have some kind of an agent or a evangelist, not only for ethics, but also for the personalization knowledge to make sure that it's all over the company and not just in one department. And that's what it would be mostly see today, a personalization experts are there in one department, but they are not mangled into all of the other touchpoints user frequency. I see that the table. Indeed, and to come back on your question, I think that's, so you said, when the data is in order, you have the people, you have the tools and everything, what else do you need? And I think you need something that actually you need upfront is a vision on how to use that, because it's only when you know why you do things and how you're going to use it, that you can have your tools in order in your data in order and everything cleaned. Data with the always serve for the vocabulary, but she didn't shit out. So having not only a lot of data, but the right data, the priorities between what's calculated, what's declared, what's any type of data, and then knowing how you want to use it is the most important for me. I think everything starts with the why, with the vision. And so what is the why of hyper-personalization, or just personalization, because like you said, Martin, not everybody needs it, so personalization, what's the why? Again, it has to bring value to the customer. If it brings value to the customer, it will bring value to careful in this case, a happy customer, a customer that finds the product that he needs will come back, will spend more, will spend more careful instead of the competitors. So it's really finding what will make it worth for the customer to come back in our stores, because in the end, we are in a business where frequency, the size of baskets are important, so it goes back to very basic things, where we have the chance to every week make the customer come back or losing as well. And I think that having that vision on allowing us to dream on longer term, but acting on very, very short term, I said every week, but some customer pass every day, every two days. So that's also an important aspect of looking far, acting very fast. Looking far and acting now. I think that's really key, because personalization is often seen as a tactic, the same as performance marketing, they see it as a short term value, but it's not. And actually, it should be something that uplifts your brand value, so you can create real customer lifetime value, you have a long term vision, a long term value for your brand. If you have a good user experience, it uplifts your brand equity. It's not only just a tactic to make sure that you have the next best buy done. Indeed, we often look at a campaign we've done, we read it. We look at how it performs, but that is really a two short term mistake, I would say, because we analyze one communication set. We send over 150 million communication every year. I think that we should look at the relation we built with the customer, rather the sentence, the email, whatever we sent. And that's why I'm coming back on the level of personalization. It's okay to sometimes being very basic to pass a message, and sometimes being very close to the customer with a high personalization. It all comes back to value and empathy. Yes, it's the same as in advertising for me. Advertising is interesting when it brings value, the same with personalization. You can see it as a tactic, or you can see it to really bring value for a customer. Wise words. No question. Any tips for people who are listening, anyone who wants to take a better look at the personalization they're doing or wants to uplift it, what should they know? I think you can start personalization tomorrow. Everybody can start doing better personalization tomorrow. The thing is don't be afraid of the big rolls roses you need to implement. It's not the case. You can start by doing simple use cases, and you can really have an impact on long term value when you start tomorrow. That's only building four use cases in a year. If you don't have the capacity, if you don't have the team, just start. I think that's the main key value, and always check on a quarterly basis. What did I do? Did we do well? Should we change our approach? Do we have a long term vision that we're building up to? That you're not only campaign focused, but really long term brand building and data foundation building on a long term? I totally agree. I think, make a first step, talk to people. The conversation will start automatically, interact on what people answer. Don't look only at what you're going to say. Listen to the feedback, the feedback is an email that is opened, call to action that is used, a coupon that comes back, and I think that every little action will make it at the end of the year. Well, it sounds like personalization is within reach for everybody. Thank you very much. Thank you, Martin. Thank you, Anthony. Very good. Thank you, Sarah.