Developing loyal customers in today's highly-competitive, omni-channel retail environment is critical. Retailers must invest in systems and technologies that enable to do this.
– [Joe] Hello and welcome to the live TotalRetail webinar, “Customer Loyalty: The Holy Grail in a Digital World.” Today’s webinar is being sponsored by Optimove, the leading customer marketing hub helping over 250 brands drive their entire customer marketing operation. Optimove combines the science of data and the art of marketing to deliver personalized, realtime multichannel customer communications at scale. I’m Joe Keenan, Executive Editor of TotalRetail and the host of today’s event.
Before we get started, let me take a second to point out the tips for attendee’s widget on your console. It’s the blue one with the wrench on it. If you missed the Tech Tips video we played leading up to the webinar, you can always click this widget for more information. Here today to discuss how brands can develop long-term relationships with their customers and maximize lifetime value, we have Pini Yakuel, the founder and CEO of Optimove, as well as Jen Gulley, Director of E-Commerce Sales at WestPoint Home, a leading retailer and supplier of high-quality bed and bath products for over 200 years. So without further ado, I’d like to pass it over to Pini to get us started.
– [Pini] Thanks a lot. Happy to be here. Good afternoon, Jen.
– [Jen] Good afternoon.
– So guys, just a little overview and a brief bio about myself. So my name is Pini Yakuel. I’m one of the founders and CEO of a company called Optimove. It’s a company that I personally set up back in 2009 in Tel Aviv, Israel. Today, I live in New York City and we are a company that’s always trying to understand what drives customer behavior and how we can create a win-win situation between brands and their customers so that customers will fall in love with their brands and choose them again and again over the competition. Today, we’re broadcasting from our Chelsea office in New York City. It’s a little bit rainy today, but other than that, not too cold and I’m really happy and excited to be here. Over to you, Jen.
– Thank you, Pini. It’s Jen Gulley and I have about 20 years of experience in e-commerce marketing. I’ve worked at 1-800-Flowers, SiriusXM Radio, Ann Taylor, and right now, I’m at WestPoint Home. And really, my focus is on making the best possible experience on the websites and kind of really driving that home and retaining the customer to get them to come back and, you know, be a great repeat customer. So I think we should just jump right into the material right now.
– Great. Sounds great. Thanks a lot, Jen.
What Is Customer Loyalty?
– So we’re going to be, kind of, like playing it together and talking about different ideas and notions. But a really good way to get started is give this notion of customer loyalty a moment. And there’s a reason we show a picture of a trophy, and this is the Holy Grail in the Digital World. I think that, you know, there’s something that’s happening in the economy in general, you know, besides talking to loyalty marketers or customer marketers. I think that in today’s economy, it’s always been important to foster customer loyalty, but I think that today, we see that customers have so much choice. And I think, in 20, 30 years ago, a lot of us, we’ve been captivated by a vendor. It still happens from time to time, probably, with the, you know, with the Verizons and then time order right where you’re in a specific zone and you have to use their cable and you don’t have a choice, so they have, kind of like, let’s call it forced loyalty. But in most cases, that’s not the case and customers have tremendous power to choose the vendor or the brand that delights them the most and caters for their needs. And I think, as time goes by, it’s no longer something that we can…it’s only nice to have. This is a necessity. This is something that… Businesses that are able to excel at customer loyalty have kind of like a tremendous win to gain.
– You know, absolutely. Right now, being at WestPoint, kind of direct-to-consumer business is very new for us, so we’re trying to become this kind of old manufacturing company and turn ourselves into this very like agile company to… You know, we want to be the choice that customers come to, not the big-box stores. So this is a real challenge to us and customer loyalty is the utmost important to us. So I think, you know, we want to give you guys the best tactics and kind of info to retain your customers and be successful.
– Great and yes… So a little bit…a short, kind of like, setting of expectation. This is, kind of like, my slide specifically, so I’m testifying something about myself. So I like to think of myself as the person who is always seeking for the truth. And I’m just saying that to talk about the fact that today, I will try to help you understand why things work as they work based on my 10-year experience in loyalty marketing and data science, and I’m not going to necessarily provide you with the, you know, quick hacks or quick tips for, kind of like, what is the best campaign to run right after the holidays or stuff like that. That’s not going to be the type of the content that will be provided. It’s more going to be, you know, the physics, the sheer physics of things because this is what I think I can provide, and Jen together with me. So, yeah.
So let’s get on to the content itself. I think that when we talk about loyalty, our tool, the one thing that we’re using to foster loyalty is, essentially, something I like to call customer marketing, some people call it retention marketing, some people call it CRM, right? It depends… However you want to define it, this is the field of marketing that is, you know, doesn’t have anything to do with media buying or with ads or with, kind of like, attracting fresh blood into our site or store. This is about the people that we already have registered, we already seen them visit our sites. This is about people that are already in our databases and our job is to try to leverage that asset into the… Essentially, it’s a growth engine.
So the connection is, is that the aim of customer marketing is essentially to foster loyalty. You use customer marketing to get loyalty. When you get loyalty, you’re able to increase customer lifetime value, LTV, which is, obviously, something that can change any business dramatically. If you’re able to tilt that number and push it up, this becomes, I think, monumental for the business. So this is just a connection customer marketing and loyalty. So moving over to the next slide, and this is something that makes my thinking a lot simpler and easier and this is something that I’ve developed throughout the years. But when you think about how to achieve the ultimate in customer marketing, right?
Start a Conversation With Your Customers
So we don’t want to do just any customer marketing, right? We want to be the best that we can be. It’s a specific art and craft that we want to master. And the one thing that makes me understand this, is when I think of the notion of starting a conversation. If you think about all the campaigns that you’ve ever read, every initiative that you do in your marketing…in your customer marketing department, it all has to do with starting a conversation with your customers, of course. So a birthday campaign is a reason to start a conversation. Essentially, we are always looking for reasons to start a conversation with our customers and not being pesky, not being pokey, kind of like, do it in an elegant fashion.
And this is the second point, not only that we want to start a conversation, we want to start an emotionally intelligent one. That means that I’m not just telling you what I want to say. I’m telling you something…you being the customer, I want to say something to a customer at the right time when that message could resonate with that customer. It makes sense for that customer to get that message at that point and we call this emotional intelligence just like, you know, with people. People with high emotional intelligence, these are people that are able to adapt and change the way they talk with different personas and types of people, right? So I know that my wife’s very angry in the morning, so I get away from her, right? And this is me displaying emotional intelligence. And people that do it really well, there’s tons of studies and, you know, research on people that do it really well, it shows that they’re very successful in life, at their careers. It’s a really super trait. It’s not only the IQ, right? It’s also the EQ, which is emotional intelligence. So we feel that brands, when they talk to their customers, they have to attain that trait. They have to be able to speak in a smart fashion, in a mostly intelligent fashion. That’s the second point.
And the third point is, obviously, you don’t want to start a conversation with your customers once in two months, right? It makes no sense. You need to start a conversation when it’s relevant and when it makes sense, so… And to sum it up, we call it…you need to do so at the speed of your customers. So are you able to move at the speed of your customers? And a lot of the points we’re going to make here is that most businesses are not able to move at the speed of their customers, they’re always lagging, they’re always behind, they’re sending one message, you know, once a month or two months. It’s not super insightful, right? Like, everybody in the audience, I think, if you would think which brands really excite you, like, which brands that you are the customer, which brands really make you feel that they’re smart, that they get you, that they, you know, looking to who you are and change their marketing as you change as a customer? So I think this notion of finding a reason to start a conversation, doing it smart enough and as fast enough, this is, for me, the ultimate in customer marketing.
Yeah, I think these are…both points are great. And one thing I’d like to say about starting a conversation, for me, it’s literally starting a conversation. Yes, we can do everything in digital, but with like our company, we’ve had customers that had our products for at least 20, 30, 40 years. I mean, which is crazy in today’s world, but since we are, kind of, a start-up within a major company, it’s just my boss and I, so I handle all the customer service. Well, I can spend a good 15, 20 minutes, literally, on the phone talking to these people who love our products. Just literally starting a conversation with them and having that emotional connection. They want to talk to someone, they don’t just want to talk to robot customer service, but, you know, they want to feel that their voice is heard.
And then, I can… You know, I’m kind, of most looping them into our system because then after that, real conversations, “Hey, do you want to sign up for our email program? Would you like to get some mobile alerts?” You know, I can get them hooked in to kind of the whole cycle of communication with them. But I think, you know, literally talking on the phone is something we forget about in the digital age, but I think, it’s one of those facets we need to focus on too, plus all the email digital kind of stuff that’s out there too.
– Yeah, absolutely. I think, you know, the personal touch is still something that, you know, we should be using. However, it’s obvious that Jen, probably, cannot call all her customers, right, and do it every day. So it’s, kind of like, a hint towards the data science part, so which…so she should use this, but she should selectively choose the right customer to talk to. And this is, of course, another communication channel which is great. So moving on to the next slide, and this is, again, the way I see the mechanics of things. So we want to be the ultimate customer marketers, right? So what does it take and where are we lacking? So what does it take to reach the ultimate in customer marketing?
What Does it Take to Become the Ultimate Customer Marketer?
So I think that if we would envision ourselves as, kind of like… I used to do some MMA when I was in my early 30s. Now I’m in my late 30s, but when I did that, I, kind of, like, connected very well with the world of martial artists. And I think that if we are all, as loyalty marketers, martial artists, we want to develop two higher traits, I would say. And these two higher traits are, essentially, precision and speed. It sounds naive and cliche, but I’ll show you exactly how this relates over to customer marketing. But, essentially, this is what we want. We want to be precise.
So when we talk to our customers, we have emotional intelligence, we talk to them about the right thing, at the right time, at the right moment, we’re very knowledgeable, we know who they are, we understand who they are, we have a lot of insights in terms of who they are as people, and the speed has to do with moving at their speed. It has to do with having lots and lots of ideas. It has to do with fostering a creative process within our marketing departments. And many things that… If we are able to push the envelope on these two higher traits, we would then have the ultimate in customer marketing.
So I think when you guys are always talking about initiatives and the things that you’re doing at your marketing department, you should always look at each…everything you do, every new initiative, we just Google those things, okay, is it helping me with precision? Is it helping you with speed? Do I need to improve here or do I need to improve there? I think for me, it makes it a lot easier, again, to think about how things work and why the world is the way they work. And, kind of like, if we are martial artists, then we need to master different techniques, right? And one technique is going to be good for speed and the other technique is going to be good for precision and maybe one technique is going to be good for both of them. And the way we want to lay out today’s presentation is by covering all these different techniques. So I’m just going to read them out loud.
So the first one is, kind of like, an org structure built for speed. So it starts with the way your marketing org is actually built and this also helps you or prevents you from being successful. Marketing technologies is another one. Single customer view is another one. Segmentation is another one. Scientific uplift measurement is another one. Multi-channel communication and catering for the customer journey. So these are all of the techniques you want to cover and I’m just going to start with the first one, which is the org built for speed.
Org Built for Speed
So what do I mean by that, org built for speed? So if you think about the way customer marketing has been, kind of like, invented, let’s say, 15 years ago, so only the enterprise could do it, right? Only very big companies could do it and they had very expensive system. So, typically, you would have, maybe, a big clunky campaign management system like a Unica or Aprimo or some… I’m talking about, like, older systems and you would have a data warehouse, and then you would have the mark on team, and then you would have a messaging tool, like an email tool and different tools. And around those tools, you’d have several different departments. You would have, you know, the campaign managers or the data guys, right, and you would have the marketers, and you would have the analysts or the data scientists.
So the way it used to work is that if you want to run a campaign, the marketing would come up with a brief of the segments that they want to run, they would send that brief to the data guys, the data guys would need to write a query to, you know, go through the databases, find the right audience. They’ll get a query, there’s a little back and forth there. Maybe they have priorities, maybe one is sick, so he’s not in today. Maybe he’s on vacation, he’s not in till next week. And then it goes to the mark on team. The mark on team has another brief that has to do with design creative. So everything comes together and a big question, you know, arises, so… I mean, what’s the time from ideation to execution?
So from the moment somebody has a good idea for a campaign, how much time does it take us to do it? Now, if it takes us two weeks, or a month, or two months, then the system will really be trained to come up with these ideas and even when they do come up with the ideas, they’ll use ones that move the needle. They got to do things that are big because it takes so much time and so much energy and so much effort. If you have a good idea, but it covers only 2,000 customers, nobody is going to care about it. Most people are going to say, “Move on. I mean, it’s too much work. We’re not going to do it.” But we feel that, you know, this old-fashioned approach which was, you know, mainly because of constraints or technologies and the way things happen, but, essentially, what you have here is you have, kind of like, an assembly, right?
So it’s just like, kind of like, a Henry Ford production line. The first day, you know, mass manufacturing assembly line that we own or even the Chipotle line where we all get our Mexican food. So, you know, one person puts on the beans, the other person puts on the guacamole, and it was pretty well for Chipotle because each task is very simple and it’s pretty quick. But here, if the data guy is sick, the entire campaign needs to wait for a week. So it makes us very… The time form ideation to execution is not fast enough and it creates a lot of…it makes us behave in a certain way. So Jen, maybe you have some thoughts on that.
– Yes, I definitely do. I’m, kind of, seeing both worlds out there. I’ve been at, you know, 1-800-flowers where we, kind of, have every single team we need to get every campaign executed. And here at WestPoint, we are an older company, but internally, we’re challenging every department to, kind of, change their cultural thinking and being like, okay, this is the new customer. How do we adapt the department so that we’re moving at a faster pace? And a lot of us are getting trained, kind of, in the techniques of, like, trying to become more efficient out there, so we’re really saying, “Okay, here’s the new customer. We’re going to help you, kind of, come out of that typical manufacturing, that long data cycle to be a lot faster.” Because you have to move fast. I mean, look at current events around you, if you’re not on top of them, you’re going to miss, you know, something good to talk about.
So you have to, like, slowly get your team to adapt and it’s not an overnight thing, definitely. It takes a little bit of time, but you definitely have to, like, slowly, kind of, get everybody on the same page as you and working towards that, you know, that speed and the accuracy. Otherwise, the campaign is just not going to be successful. So that’s just, kind of, my experience with it. So it’s, kind of, interesting just being the smaller company right now and doing this because we’re doing it from, you know, the ground up, so we get to, kind of, put our say on it and not just the factory is like, okay, go to the data, go to the marketing, go to graphics. You know, we’re doing it on…making it the way we want to. So…
– Cool. Yeah, I think that… So to sum this up, I think that the way you’re marketing org gets built, is definitely going to determine how you do customer marketing. And if you want to build your org for speed, the solution is that you can’t have too many pieces in the assembly line, right? So of course, designers will be… You’re going to have the people who build the content, the designers and the graphic designers and people who do copy, and then you’re going to have the CRM…I call them the CRM A-team. So the CRM A-team, they’re using technology and they’re able to both come up with the briefs, do the segments on their own, you know, come up with the designing the task, and also doing the analysis when the campaign is done, doing the data scientists, they do everything. So it’s a two-department layout. So it’s the CRM A-team together with the designers that do the copy of the content itself and that’s it. Instead of like a four or five-department layout that they had before, which is a lot slower. If you think about start-ups, so just like as Jen was saying, so she is now, you know, working in a start-up within a big company. So when you have to coordinate with yourself, it’s very easy.
– It’s very easy. There’s no miscommunication, at least for most people.
– Oh, yes.
– Yeah. So for most people, that works and I think, that’s part of… So I think, the good news for CRM talent and for CRM practitioners is that you guys are going to be, you know, front stage. It means you need to know more, but there’s so much great tools, and technologies, and capabilities today that it’s not such a big deal and with just a few more steps you guys can shorten the cycle and have a very quick layout. So remember, time from ideation to execution, I want it to be very short. I want it to be, you know, minutes or hours or, you know, for complicated stuff, I want it to be days, but not more than that. Because this creates a department who’s, kind of like, you know, fostering creativity. Every day, somebody else comes to the office, “You know what? Let’s do this and that. Let’s do… I have another idea. No, I have a… ,” and somebody’s “Okay, let’s test it. It’s not a big deal. It will take us hours or days, but we can test it.” When you have that, you have a lot of people coming up with ideas, you have…and then you cater to your customer. Then you are able to start a conversation with your customer, you have more reasons to start a conversation, smarter reasons to start a conversation because you have a creative department. So this is the first technique which helps you with speed.
So this one helps you with speed. Let’s go to the next one. So the next technique also helps you with speed and this is about marketing technologies. So again, if we go back, as I said before, 15, 20 years ago, we would have this on-premise technologies, was very, very expensive. Only very big companies were able to afford it. It was pretty clunky. We didn’t have the revolution of the user experience and, you know, proper tools that we have today. So marketing technologies, first and foremost, they give us the magic of cloud computing, API connectivity, automation, and data science.
So we’re getting all of these wonderful things for so many tools and I can tell you, as somebody who owns a company, you know, that produces software, you guys are getting it pretty cheap. I mean, the software world is highly competitive and highly fast, and this pushes prices down. So as years go by, you’re always getting better and better technologies for lower, lower price, and as long as you just leverage it and utilize it, you’ll get amazing ROIs on these tools. So you can accomplish so much more. Another thing that happens with a good marketing technology is that, essentially, it wraps and productizes an untitled topic.
So if you’re talking about customer marketing, I want to start doing customer marketing in my company, and I haven’t done it before, I know by this system, that system is going to be my gateway to this field, right? I’m going to have all of my data there, I’m going to have all of my campaigns, the communications with customers, I’m going to have my results, my stats, my metrics, I can show it to my managers, I can… This becomes, kind of like, you know, it wraps everything in the box and it makes it very easy to manage. And the last, but not least, when you get a good technology that comes with good methodology, it enforces work methodology. So in many cases, maybe you have a few people working in the company and some of them are new and they’re not up to speed with how you should do certain things, but if you have a good tool, that tool would enforce methodology, that tool would align everybody to work in a specific way that has been proven to be the best way in that specific field. Cool. So this is the second technique we want to master, which is marketing technology, all right? The first one was an org built for speed. Let’s go for the next one and the next one is a technique that has to do with precision. Not with speed, with precision.
Single Customer View
So when we talk about precision, we start talking about knowing the customer, right? And for us to be precise when executing customer marketing, the key here is intelligence, right? We want to collect intelligence, collect data, understand more and more and more about our customer. Obviously, there’s tons of ways to know our customers. We can, you know, invite them over for a panel and ask them some questions, but that’s going to be a small sample. Today, our customers are telling us who they are in the data. They’re showing us who they are. We know what they bought, we know that they visit our website, we know which pages they looked at, we know if they open our emails, we know if, you know, if you put on some surveys and they fill it up, so… And sometimes, we can get additional data about what they do while in the web, you know, like on social media and things like that.
So we have a lot of information on our customer and the very first thing you need to do, and this is in the third bullet, this is a… I think it’s a basic necessity in order to practice customer marketing. It’s this notion of a single customer view. I like to cook a lot and I think, you know, just like when you cook, if you have a proper pantry, right, that’s nicely stacked up, you can start cooking. So this is the single customer view. It’s, kind of like, your pantry as a customer marketer. You need this to get started. This is the foundation of everything, right? This is the cornerstone. So the problem is, is that why is it so difficult to get to a single customer view? Because customer data is dispersed across multiple systems or databases, right? So in my billing system, I have the customer payments. In my CRM system, I have customer, you know, support requests and tickets and stuff like that. In another system, I have…in my website vendor, I have like, “Oh, they’re proud of me.”
So there’s various systems, each one of them captures a small file of data, but, essentially, it’s the same person. So when the person, you know, is there, they expect us to know everything. So a single customer view, you see down there, it’s, essentially, a flat file. So you get a customer ID and you get multiple columns coming in from different systems and each column represents, you know, a piece of data, something about the customer. So like how much they’re spending, their gender, their state, you know, the level of affluence, many things that we can know about our customers, we’re going to represent just like a very wide Excel spreadsheet because that’s the easiest to understand. I can just read one line and I have everything I want about my customers.
So this is the single customer view and, kind of, like bringing into a flat structure. Some of you, probably, have it in your company, some of you don’t, some of you, you have it partially, which is also okay. But basically, I think that this is a very strong foundation. I would say this is where I would get started to know my customers. Okay. Great. So after we have our single customer view, we can start cooking, right? Now we have a full pantry, now let’s start cooking, and probably the first dish we’re going to cook… Actually, I have here also pasta sauces, but it has nothing to do with us cooking. I’ll get to it in a second, but probably, the first dish and the most important meal for us as customer marketers is something called segmentation, right?
So segmentation is probably something that you guys have been hearing about for years, you probably have different notions in terms of what it means, maybe it’s just segmenting customers by gender or by state, right? Is this segmentation? So the answer is yes. But segmentation is a very powerful and a very broad notion that we need to understand. So this will be a big… A big chunk of our presentation is gonna be about segmentation, and the reason you have all of these Prego pasta sauces is… I think this is the best way to explain the miracles and benefits of segmentation. So here’s a little bit of a story, and you can find it in the TED Talk by Malcolm Gladwell. So basically, I think in late ’70s, supermarkets were a lot different than today. Today, when you go into a supermarket, you go look at the pasta sauces, there’s like, you know, 100…you can lose yourself and not know what to buy from all the varieties, different types of sauces. Same goes for cereal and different types of food, right? So why is that? Why are we living in a world that we come to the supermarket and there’s so many different types of pasta sauces? Right?
So this is an important piece of history. So back in the late ’70s, these companies like Prego, before they launch a new product, they do a lot of testing, right? They get focus groups and people are…they do samples and people taste it and they say what they like most and what they like less. And the people, at the time, everybody was looking for the perfect dish, the perfect formula. Give us the perfect pasta sauce that everybody on the planet is going to fall in love with. And the problem is that there’s no such thing. There’s no perfect dish for all of humanity, right? There is a lot of clusters or a lot of different segments of people with different palates, and different tastes, and different taste buds and different probably, God knows, you know, childhood preferences, and different reasons why we like different types of sauces.
So the first person to do this, his name is Howard Moscowitz. He’s a Harvard Ph.D. researcher and what he was doing, he was conducting this analysis for Campbell’s sauce, basically, for the brand Prego. And what he did was instead of testing, you know, one recipe or four different recipes, he was using different variations. He was using, you know, the level of spiciness and how much garlic is in it and the consistency of the sauce, if it’s more liquidy or more chunky, with tomato. So he was trying and he came up with 45 different recipes and he was bringing a lot of people to taste it. And after he got the result of it, everybody ranked everything that they liked, he has this technique where he can find clusters, and he found, at the time, it was three main clusters of what people liked, and back in the late ’70s, Prego launched…instead of launching one jar of pasta sauce, they launched three, and at the time, they were not doing so well and they took over the market. They immediately became number one and the rest is history.
The rest is history. And I think this… I mean for me, it means, you know, when I was studying this notion of segmentation, again, trying to explain to myself, I think this story really resonated with me because it’s about foods and I think it makes so much sense why this works. Because we are different as people. So we need to find these personas and then talk to them differently, cater to them differently, provide them with a different offering.
– Yeah. I think also with like segmentation and data and stuff, I think it can become easily overwhelming and you can get just crazy because you have data coming in from all directions these days. And so I think, I know we have all kinds of different, you know…people on the phone right now from small to large companies, but I think, you have to start small with your segmentation. The groups, like, start testing. You don’t have to figure out every single little one out today. But start with some of the bigger ones because they’re going to drive the most revenue. And then slowly, you can, kind of, break down each of those groups, but just kind of start with some high-level rules of what you want to do because I know personally, like, the data, to me, it becomes overwhelming and, you know, you may get flustered where…should I start with this group or that group? But make some just high-level goals like I think this group’s going to buy, like, our blankets, in my case, you know, or they’re going to buy the blue blankets, you know, or the expensive blue blankets.” So kind of just like start smaller with your groups and stuff because, you know, 45 different groups becomes overwhelming with tomato sauce especially, but, you know, you’ll narrow it down to your key groups eventually.
– Absolutely. I concur and I think it’s absolutely wise… By the way, he did eventually create three, because also probably, for the company it wasn’t that easy but… Well, yeah, I think it’s… I think Jen is 100% correct and basically, I call it… I think today, we talked about having so much data, but to some degree, I think we even have data obesity in some ways. And our job is to curate, right, is to actually get focused, find something. And I agree 100% with Jen. The philosophy and…by the way, we have a few, here, examples of simple segmentation that can get you started. This is one of the things. So great. So we understand the pasta sauces and why segmentation is important. And now, let’s move to some techniques and how it works.
So basically, the world of segmentation, I know this looks a bit mathematical, but just to explain one important notion, we have two types of segmentation, and it’s important for us to understand that segmentation is something that we do for the purpose of organizing and summarizing our data. This is the purpose. There’s no right answer, there’s no wrong answer, it’s about organizing and summarizing our data and the…especially, if you’re not trying to build it for a specific like… I’m not going to tell you, but in the world of machine learning, you have supervised learning and unsupervised, but typically segmentation is unsupervised. So a lot of people, what they do, they do rule-based segmentation. So if I have this, this is a two-dimensional chart. Let’s say, I have a monetary value on the y-axis and frequency on the x-axis, and I have all these dots of different customers, so I could divide my database into four different clusters, right? People at the bottom left, they could be people with low monetary value and low-frequency value. And then the top right, it’s going to be… Probably, it’s not going to look exactly like this, but usually, we’ll say people that bought more above or below $1,000, and above or below four visits, right, or four purchases.
This is the rule-based segmentation, which is pretty common. It’s the one that most companies use and it’s a very good start, it’s a very good start. I would do some analysis in terms of where to put those, you know, those specific borders in the data. But the next type of segmentation, it’s called cluster analysis. In this type, we’re actually letting the data tell us how the segments look like. So it’s kind of like looking for, in this case, it’s two-dimensional clouds of the data. So where does the data centralize? How does the data reside? And in this case, I see, again, four clusters, but I see it’s laid out completely differently. So customers who belong to the red cluster, you know, in the rule-based scenario, they’ll be divided into two different segments. So these are kind of like the two, I would say, meta techniques that are being used for segmentation. Let’s see a few examples. So Jen was talking about starting easy, so probably the best retail segmentation that… The first one you should do if you don’t have anything else is something called RFM.
So RFM is the Recency, Frequency, Monetary. So if I look at these three attributes for a customer, there’s a good chance I’m going to know who are my best customers, who are my second-best customers, and things like that. So monetary is how much money are my customers spending. I can do it ever since I met them, I can use monetary value for the last year, depends on what you want to achieve. Frequency is how many times they’ve visited ever since I met them or in the last year or the last six months depending how you want to treat your customers. And recency could be days since last purchase, or weeks since last purchase, or months since last purchase, it doesn’t matter.
And the simple way to do it is we can divide each one of these attributes into, in this case, it’s into a quartile or decile, or it doesn’t matter, but it means that we’re going to use rule-based segmentation on each one of these attributes and cut it even, in this case, at the 75th percentile, 50th percentile, 25th…so we have four quartiles or segments. Same goes for frequency and same goes for monetary. And then I can say, you know, people that…in frequency, I’m looking for a low number, right? A smaller number is better, so is recency. In recency, a smaller number is better and obviously, in frequency and monetary, a bigger number is better. But if somebody is in the best tier of R, the best tier of F, and the best tier of M, these are probably very good customers. So not only that they spend a lot of money and that they do so with lots and lots of visits, will probably buy this… By the way, Jen, it’s really…it’s always an interesting question. What would you prefer? So would you prefer a client that bought like one order for $1,500 or three orders of $500?
– Three orders for $500. Because I can market to them again with different things to keep coming back and spending over and over.
– Right. So it’s probably good indication that somebody chose us three times.
– Right. So it’s probably more likely… So if you try to predict, let’s say, the future value, they probably have better future value. So this is what… So yeah, it’s an interesting question. So because, obviously, by the way, monetary and frequency are very much correlated, right? So in many, many cases, you see when this is high, that is high. So it’s interesting to look at those groups of low monetary value and high frequency or very high monetary value and low frequency, right? This could tell us about opportunities. But, essentially, this is a very important segmentation and yes, just like Jen was saying, RFM, I can tell you, it’s a very strong predictor for future value.
So probably, the customers that spend a lot and visited a lot, if you wait a year or two years, these are the customers that will continue to spend a lot and continue to visit a lot, obviously not everyone. But as a cohort, as a big group, they will outshine other groups in the RFM which are not as strong. So this is definitely our RFM analysis. Another kind of like RFM analysis that’s done using cluster analysis… So hopefully, you guys can see. So in cluster analysis… By the way, this is a segmentation done on monetary and frequencies, so you can see net paid amount and number of paid orders.
So I have two attributes. But you see that I have four clusters…sorry, three clusters with exactly one order and two clusters of two orders and the clusters of one order have different spent. And you can see I have this cluster here with $100 of spend in one order, and at the top, I have two orders in $118. So the cluster analysis, actually, organizes the data in a way that I didn’t know existed. I would not know to do this division myself. I would not know how to draw the line like this, but this tells me and you can see also the number of customers. So the second cluster, there’s only 79 customers who present this behavior of buying two times, but for a lot lower amounts. So cluster analysis doesn’t care what you have pre-decided, right? Cluster analysis segments as the data is laid out. So in many cases, it’s going to give us a lot more insights. Okay, moving on. So lifecycle stage segmentation.
Customer Lifecycle Segmentation
So after we do RFM, RFM is pretty cool, there’s another segmentation that’s very important and very much top-level. It’s the lifecycle segmentation. And with the lifecycle segmentation, we have the…typically, we have new customers, active customers, and customers who churned. The reason it’s super important is because we understand as marketers is that for new customers, you want to incubate them, right? They’re just coming in, they’re charging up for the first time, and we want to understand who they are as customers in the best possible way, customers who churned or if you want to win them back. So it sounds like understanding this different marketing philosophy to apply for each one of this. So how do you do this segmentation?
So let’s start with new customers. So new customers, typically, it’s time-stamped the first purchase, right, or time since registration, depending how you want to define a user. For example, I can tell you that three months is a very good rule of thumb in e-commerce to define as the period for new customers. So in the first three months from the first purchase, right? How do we choose this thing? We try to see how likely are they to make a second purchase. So we want to give them enough time to make a second purchase, but on the other hand, we still want it to be short so they would be considered as new. We’re not going to give them two years and keep them new customers for two years. Makes no sense, right? So when we analyze the probability of that cohort of people buying more than once, buying again, we see that in the first three months, we see most of the behaviors and then after three months, in the fourth month, it starts to stabilize, so we are getting the best benefit in the first three months.
So I would definitely urge you guys… I’m starting to speed up a little bit because I see we don’t have a ton of time. I would definitely urge you guys to kind of like divide your database into lifecycle stages and look at that new customer lifecycle state. Adapt your strategy for new customers, right? Talk to your new customers after their first purchase, you know, talk to very strong new customers versus the regular new customers, right? So it’s a very important kind of like stage to look at. Another important stage is churn. So churn is mainly based on recency. So I don’t know, in most businesses the customer is not calling me and saying, “Hey, I’m never coming back.” Right? Unless it’s the bank or a telco company, mostly with retailers, they don’t do…it’s a silent churn, right? We don’t know that they churned.
So looking at recency, if Jen… If I have a store and Jen didn’t buy for six months, depending on the frequency of that store, but let’s say I’m selling groceries, if Jen didn’t buy for six months even probably two months, that’s a very bad sign. So there’s a very good chance she’s not coming back. So we can send people to churn based on recency. So if they didn’t buy for nine months, six months, you know, a year depending on the business, I’m going to send them over to churn, and we can also do something we call a churn factor where we’re going to do it, you know, on a personal basis per customer.
So for example, if Jen’s buying every two weeks, she hasn’t bought for two months, her churn factor is four. So I’m used to seeing her every two weeks, but she hasn’t bought for two months, so it’s four times her average frequency. But for another customer who’s buying once a month, for him, churn factor of four, it’s four months, right, if he buys once a month. So churn factor is a nice way to kind of like create a churn criteria which is personal and per customer. Okay. Next lifecycle stage is reactivated.
So reactivated, it’s a pretty important stage. A lot of people don’t use it because if you only do new, active, and churn, once people come back from churn to active, it’s a problem because a lot of them churn again. What we have found is that if you put kind of like a middle ground, another stage where you can call it back from churn or reactivated, you let them sit there for another three months, right? So the same period that you have for new, right? So if you had for new three months, so after coming back, let them sit there for three months. Let’s see what happens? Let’s see if they zigzag and go back to churn, which will happen in many cases, or let’s see if they can like stick around and become active.
A few things we know about reactivated, we know that these customers behave just like new customers, so we need to re-incubate them all over again. You can see in their future value, these groups are pretty close. It’s roughly the same thing. We also know that if a customer reactivated on his own, he has a better future value, so it’s not like a cherry picker that’s looking for an offer. And we also know that if a customer churned only once before, there’s a good chance, so we need to work harder because if they churned a lot of times before, they do know how to zigzag. So they’ll come back, they’ll go to churn again, then they’ll come back again. So we need to see how many times did they churn before. Moving on. So this whole process can be combined and this is what we do at Optimove. We combine this notion of stages, segmentation layers, the micro-segment into this very kind of like a whole customer marketing approach that goes all the way down to micro segments. But you guys can definitely stick with the stages, and RFM, and things like that. And lastly VIPs. This is a very strong segment to cater for because… We all know the Pareto principle, right?
The VIP Segment
So we all know that, you know, probably 80% of the clothes I’m wearing…sorry, 80% of the time, I’m using 20% of my closet, right? And the rest of the 80% is a waste. And in the radio station, 20% of those songs are being played 80% of the time, right? The same go with your customer, right? There’s a small, small segment of your customers, in which some retailers it’s a lot stronger than 20/80 that actually generates really high value. And these people, you know, you should treat them like kings. Because eventually, if you just lose 2, or 3, or, you know, 20 of them, depends on the business, it could mean a lot of impact to the company.
So definitely, create a separate segment for VIPs. These are customers in a cluster of their own. There’s many ways to find them, right? You can do something simple like the top x percent, like top 5% of my customers. So just sort your customers by spend and take the top 5% or top 2%. Or say, I have a budget that I can treat 1,000 VIPs, no more, so just take the top 1,000, right? And we have other scientific ways to kind of like find them and there’s more precise ways to find them. So you can use either the data science ways or the simple ways. And the ROI of segmentation, again, it comes just like in the past of those example, if you’re running campaigns in all of your customers, maybe you’re going to get an uplift of $20 on average, but if you’re dividing it to many, many segments, in each one of these small segments, you’re going to have a different message, you’re going to get a different uplift, and on average you’re going to hit a higher number. That’s the idea.
Scientific Uplift Measurement
The idea is exactly the same as in the past example. Okay, before we wrap-up, I want to talk about two more things. One of them is scientific uplift measurement. So essentially, when we run a campaign, what we want to do, we want to use this thing called the control groups. So if I’m running a campaign for 1,000 people, I want to isolate 100, set them aside with as a control group, and if this campaign is running for 7 days, I want to wait for 7 days, see the actual response rate of the control groups versus the test group, and I want to see that I’m able to beat it. So if I’m doing 20% versus 14% and my average order value in the test group is 200 versus 150, then I know that I made money. A lot of people are not measuring their campaigns in the proper way. You’re maybe looking at, you know, open rates or click rate or looking at revenue attribution, but you’re not attributing what would have happened without you running the campaign.
If you’re targeting your VIPs, they’re coming to your store any week, regardless of any campaign, right? So you always have to gauge the uplift just like the clinical trial. So I can always going to give you, if they invent a new drug for headaches, they’ll give you a tic-tac to see if the placebo effect is actually kicking in and not the drug. And cool.
Catering for the Customer Journey
So last technique is catering for the customer journey. So this one helps us both with speed and precision. You know this notion of a journey, it’s really good for new customers and reactivated customers. Right? So we want to take them through this process of, hey, here’s a welcome email. After two days, if you bought, here’s kind of like a coupon for the next purchase or if you didn’t open the email, I’m going to send you something else.
So we’re building this layout and the one thing I want to say, remember, when do you want to start a conversation? When do you have a good reason to start a conversation? So we call this the static journey. And then the infinite journey, which is the second type, is instead of us blueprinting everything that could happen, we basically build different segments or different states that the customer could be at. Whenever the customer reaches that state, we’re going to serve a specific message and then the customer would change. Again, once they change, we serve another method. So we let each customer draw their own journey based on how they’re changing the data versus us kind of like blueprinting what would happen. The good thing about the blueprint is that you can show it to other people, it’s very easy to understand what happens, but sometimes, a very important thing with your customer could have happened and you forgot to cater for it. Also becomes very hard to manage these things at scale.
So this is the static journey versus the infinite journey. And last, but not least, is the multi-channel communication, where when we do these campaigns, this starting a conversation with our customer, we want to make sure that we do it at the right channel, we want to make sure that we sometimes reinforce the same message over a few channels because then our primitive brain is going to tell us, “Oh, this is something I need to look at,” right? “I can’t ignore it.” So it works because it caters to how we are wired. This is how our brain works. And sometimes we can complement between two messages and say, you know, I’ll send you a text because I know that there are 100% open rate, and then I’ll tell you in the text that there’s present waiting for you at your email, then I’m going to use the rich template that I have with email. So I was kind of like running, running very fast because I noticed that we didn’t have a lot of time left. So now, it’s a good time for Q and A. Please take it away, Joe.
– Thanks, Pini. So I’m going to jump in here. So we do have a couple minutes for your questions. As noted on your screen, you can do so by submitting your questions using the Q and A box on your console. So my first question, I’ll open this to both of you, Pini and Jen as well.
The question is, “Do you have any thoughts on how you can predict future lifetime value of a customer?”
– Yes, and to predict future lifetime value, we talked about kind of like the RFM as being a very good predictor and that’s a good way. Another way to predict… It depends on the means that you have, right? You can go fully blown with data science and build like a very sophisticated model just like we do in Optimove. But you can also do it in a more simple manner. You know, you can look at it kind of like a cohort of accumulated revenue and then kind of like use even Excel to draw the line how it’s supposed to continue and get it from there. So those are the few ways to do that depending on how much you want to invest, but I definitely recommend kind of like managing that number on an ongoing basis.
– Okay. My next question I have here is…the attendee asked about reconnecting with loyal customers again after a period of no contact or inactivity. Thoughts about techniques there and then…yeah, so around kind of reconnecting after a period of inaction from some of your best customers.
– Yeah, so…
– I mean the technical and the fun data side of it, I think, just from a pure marketing side, I think you have to really look at what they were buying and almost think outside the box, something that you maybe can send them or offer them that kind of going to grab their attention. I know at Flowers one time, we actually send people, I think, it was like two to three cookies in a little gift box just kind of…just a little surprise in the mail to try and get them to re-engage with us. Because, you know, who doesn’t love a little sweet treat in their mail that shows up? But I think you just have to think outside of the box. And so like slamming them with another email or another text, I think you just have to be a little bit unique there and I think that’s where the data is so important to see, kind of, what were they buying, you know, and how much were they spending to kind of cater it to them. That would be my suggestion on that.
– Cool. And then the next question I have here is, “Do you have some ideas or thoughts around tools that can pull data from multiple CRM to that one system to give you a single customer view?
– Yes. So typically… This question comes from Brian. So typically, Brian, the way this works is that you need to… It’s around ETL processes, right? So you need to extract, transform, and load the data. So that process of uniting data from various sources. So then it depends. If it’s kind of like a Google Analytics, Gmail, and Dropbox, and stuff like that, then you have some tools and connectors to these things, but typically, data sits within more enterprise systems and then it becomes slightly more murky, you need to use the…
So we at Optimove, we do it for all of our customers. We build the system for them in our cloud. But what we do is our data scientist just connects to all of these places, builds the pipe, you know, gets all the data, cleanses the data, this…you need to do some sanity checks, you need to do, you know, you need to kind of like look for anomalies because… The very important thing about data, it has to be super clean, it’s garbage in, garbage out. I don’t know any like magic solutions, but definitely, there are some agencies or consultancies or even in-house, if you have a dedicated data engineer, probably he can pull it off. There’s many tools that it’s not that hard, but definitely, you need somebody with their fullest attention.
– I’ll try to get one more in here. For you, Pini and Jen, are you using third-party data, you know, and also social data as you look to… Are you integrating into that to your first-party customer data to build that single view of the customer? Maybe some thoughts there and an example?
– So we mainly rely on first-party data. We do append… You know, sometimes you can learn a lot from a person’s zip code, so you can know a lot about…there’s a lot of statistics from the IRS or even from the Census about, you know, the level of affluence and things like that. I found that in many, many times, just looking at a person’s, you know, website activity, like the types of products they look at, you can know if they’re a female or male. Based on the types of products they look at, you can know if the products are very expensive, or if they make a purchase you can know about the level of affluence. In many cases, better than third-party, you know, solutions. Definitely, DMPs, if you, you know, if you live in the acquisition world and you want to come look for customers who are, you know… If you were selling cars and you get people that looked at a website that compares prices of cars, then definitely, they’re on the market to buy cars. I’ll say it’s a good timing to approach them. But in terms of getting to know the customer, I found that first-party data is the most reliable for me.
– Okay. Well, it looks like we’re just about out of time for today. I want to take this opportunity, on behalf of TotalRetail and Optimove, who was the sponsor of today’s webinar, to thank you for attending. Be sure to check out our webinar page to get information on all of our archived and upcoming webinars. If you would take just a minute to fill out the brief feedback survey that will appear on your screen next, we would very much appreciate it. Your feedback will influence the webinars we bring to you in the future. So with that, I want to say thank you and I hope to see you all at the next TotalRetail webinar. Have a good day, everyone.