Kirill: This is episode number 97 with Digital eCommerce Specialist Nick Pape.
(background music plays)
Welcome to the SuperDataScience podcast. My name is Kirill Eremenko, data science coach and lifestyle entrepreneur. And each week we bring you inspiring people and ideas to help you build your successful career in data science. Thanks for being here today and now let’s make the complex simple.
(background music plays)
Welcome back everybody to the SuperDataScience podcast. Today I’ve got an exciting episode with an interesting guest, Nicholas Pape. He used to live in the UK, now he’s back in Australia, and he’s an eCommerce Specialist. So, first things first, this episode is available in video format, so head on over to www.www.superdatascience.com/97 if you’re in front of your computer, and you can watch it there on video. And in today’s episode, you will find out a lot of insights into the world of eCommerce and selling retail products online. So Nicholas has about 20 years of experience in the space of eCommerce, and now his strategy about data science is to learn data science, to use data science techniques, in the space of eCommerce. It’s really interesting. So he’s not necessarily looking to become a full time data scientist, even though that might happen as well, but his main focus is to borrow those most important, most powerful, methodologies to bring into his domain. So this episode could be very interesting for those of you who are in similar situations.
And in this episode we’ll talk about eCommerce, CMS and customer management systems, geo-demographic segmentation, we’ll discuss some interesting insights about purchasing habits of people, such as looking at emails on Friday, purchasing on Sunday on eBay, or why purchases drop off in retail usually in February. We’ll talk about seasonality, we’ll talk about the customer experience. So lots and lots of interesting topics that we’re going to discuss, and I think this episode is important because it helps put a face to a name, so to speak, in data science. A lot of the time, we’re working with data, but we don’t think about the customers who are behind that data. This episode will help you keep in touch with what’s actually going on and better understand what the customers are going through in this whole process.
And without further ado, I bring to you Nick Pape.
(background music plays)
Welcome everybody to the SuperDataScience podcast. Today I’ve got a very exciting and fun and interesting guest, my friend from England – actually from Australia, but he lived in England – Nick Pape. Nick, how are you going, man?
Nick: Good, Kirill. How are you?
Kirill: Yeah, good, good. So it’s been a while since we met.
Nick: Yeah, about a month and a half, yeah.
Kirill: Yeah, only. So for those who don’t know, we were travelling on a trip with Hadelin, a Euro trip, and we were having catch ups, and Nick was at our catch up in London, and dude, we had the best time. You were sitting next to me during dinner, we were just chatting away all night long, right?
Nick: Yeah. It got pretty deep fast. It was good. Good banter!
Kirill: Yeah, totally. So I really appreciate meeting aussies. How did you get to England, what are you doing there, man?
Nick: I came over about 8 years ago to get a digital education, basically. I got sponsored, had a great role, and thought yep, UK is where it’s at to learn digital and eCommerce, and that’s what I’ve been doing for about 8 years, yeah.
Kirill: Yeah, and how’s the weather?
Nick: Stereotypically true!
Kirill: Stereotypically true. Yeah. We were there, the first day, it was so sunny and nice, I was like, “all these rumours about London, they’re not true, this is amazing.” And then for the remaining 2 or 3 days we were there, it rained 5 times per day! It was crazy, man!
Nick: Yeah, you get every season in one day, even in summer!
Kirill: The question is, why does somebody trade the sunshine of 364 days per year in Australia for what you get in London? Why would you do this to yourself?
Nick: I think I’ve been asked that question every day that I’ve lived here! No easy answer. I think it’s because you actually come over here to get out of here, to get to Europe, to say it quickly.
Kirill: Yeah, yeah, true. And you can to Europe in like a 2 hour flight, right? Anywhere in Europe, maximum.
Nick: Yeah. You can get to Amsterdam in 40 minutes. So that’s pretty cool. You can’t do that in Australia!
Kirill: Yeah, and a lot of the time, people would say, “oh, you’re in London, but you’re going to Prague or something, that’s so far away.” No, it’s not! It’s like 2 hours!
Nick: I’ll tell you what, when you visit Australia, you start talking to your friends, and they say, “where do you travel,” and they’ll say, “oh yeah, we went camping here.” And they’ll say “oh, where do you travel, Nick?” And it’s like, “oh, well, I’ve been to Italy this week, France that week, Spain that week.” And then it just looks weird.
Kirill: Alright, data science, right? So wondering, what are these two idiots chatting about here? We want to talk about data science, right? Tell us about your story. You are in CRM, is that correct?
Nick: Yeah, that’s it. Newly into the world of CRM, so it’s all about looking at data and the insights from the data to then make action on that data largely to help a customer achieve a goal, or to find new customers or reactivate customers. It’s really interesting.
Kirill: And CRM stands for Customer Relationship Management, right?
Nick: That’s it.
Kirill: Gotcha. Okay, so you’re looking to get into the space of data science, or you want to acquire some skills from data science to apply in your CRM role?
Nick: Yeah, it’s the latter, so definitely get skills. My job, whilst it’s a CRM role, is largely eCommerce. So it’s not just about learning about the customer journey and the life cycles of the customer, but it’s also doing everything else, you know, what channels we look at to get more of those customers. So in order to do that, things like Excel only get you so far. I’ve had to quickly upskill to start delving into how I start analysing that data quite quickly. And now it’s just by pure osmosis (and I did some of your courses), I’m starting to explore this potential data science arena, which is a very new world.
Kirill: Okay, cool. So how is it going? What about the tools? What have you picked up so far that you’d say is the most valuable for your CRM role?
Nick: I came in with just really good Excel/Power BI experience, and then quickly learned SQL. And then somebody said, “Hey, have you ever heard of Tableau?” I said no. And then within about 72 hours, I was like, “This is the best thing since sliced bread.”
Kirill: (Laughs) Since what? Sliced bread?
Nick: Sliced bread, yeah.
Kirill: Yeah, I’ve heard that one.
Nick: And then coincidentally—I’m not plugging you too much here, but I took one of your courses where I was like, “This is actually quite easy, as a duck to water.” And by the end of the week, I was doing all these visualizations and taking people on this journey about what our customer was doing.
Kirill: So you were actually using Tableau at work?
Nick: Yeah. I got a license. I just got one of those trial licenses and then uploaded some data and it was awesome. It completely redefined the way that I was talking about what was happening to the customer and I was actually able to visualize it quite quickly, within a week I turned around some pretty cool skills and brought a lot of my stakeholders along for the journey as well.
Kirill: I learned an interesting term from a friend about I think a year and a half ago about this, it’s called “threshold concept.” It’s like, once you learn something, you never look at the world the same again. Now that you know how easy it is to create those dashboards and use Tableau, you are never going to look at the problem in the same way. You are never going to go like, “How do I apply Excel or anything?” You’re always going to know, “Wow! It can be so easy to use Tableau.” That’s how I would characterize learning Tableau and pretty much any data science tool, but especially Tableau, because it’s so powerful and you can learn it so quickly.
Nick: Yeah. I mean, I couldn’t tell you. I used to play with paid search, like Google AdWords. I used to look at that data in Excel and my computer would just churn and churn and churn. And then with Tableau, you just upload this flat data, like a text file, and then within about two seconds it has already rendered all the data and you’ve got all this insight. It’s just made everything very productive quite quickly, so it’s quite cool.
Kirill: Yeah, fantastic. So you’re moving back to Australia now, right?
Nick: Yeah, I’m wrapping up the experience and going back to a market that’s growing quite quickly in eCommerce, but it’s obviously quite a smaller country compared to the U.K. So, yeah, I’m hoping to capitalize on that and then go forward with this learning and seeing if data science is going to be a new feather in my cap. Because I’m not a typical analyst, I’m not a mathematician, so it’s quite a new world, because when you start looking at data science and whatnot, typically that’s the benchmark a lot of people I’ve observed come from.
But I think there’s a big path there for people that work in digital or eCommerce roles. You need to have this data-driven insight, otherwise it’s just subjective thought. It’s a good skill that I think everybody needs to have.
Kirill: Gotcha. So you’re going back to Australia. You’re still with the same company or have you’ve found yourself a new role? What’s the goal there?
Nick: Yeah, I’m looking at a new role because the company I’m leaving is largely U.K. business. They’re a global footwear brand, but their hub is in the U.K. There’s massive opportunity in Australia. There’s lots of other brands there that I’m sure can benefit from the experience.
Kirill: Yeah, for sure. That’s really cool. You did mention that you’re going to explore, but what’s your gut feeling? Are you going to jump into data science at some point? Or are you going to slowly get the feel for it, what’s going on there?
Nick: You know, I’ve been listening to a few of the podcasts and from our chat, and I think I’ve got to put some time into Upwork, or one of those ad hoc consultancies, just to see whether there’s proof in the pudding here. Because when you’re doing things remotely it might not work as well. But I definitely want to explore it and maybe go down a consulting path.
Kirill: Yeah, definitely. That’s a good approach: Use Upwork to see if there’s proof in the pudding and to put a feather in your hat. (Laughs)
Nick: Cap.
Kirill: Cap! Feather in the cap.
Nick: I’ll try and scale back on my—
Kirill: No, no, no. It’s the best thing since sliced bread, man. (Laughs) This is so cool.
Nick: Just for everybody out here, you were loving all of these when we went to dinner, right? Remember that?
Kirill: Yeah, I was like, “Where are you coming up with this stuff?” I don’t know, is this Australian or is this British?
Nick: It’s hybrid, mate. It’s a hybrid.
Kirill: Yeah. Nick was telling me before the podcast that there’s Aussies here, there’s British people here, and Nick is somewhere here in-between. He’s got a world of his own.
Nick: Yeah. Just like you, mate.
Kirill: Yeah, I guess so.
Nick: Just a bit more straight and narrow.
Kirill: (Laughs) Yeah, gotcha. Okay, cool, so you’re going back to Aus. And I like that concept, you know, a lot of people use Upwork to start into freelancing, to become consultants, but why not? Why not use Upwork not to become a freelancer and consultant, but to test out if there is demand for these skills. You know, maybe then you will become a freelancer, or maybe then you’ll go venture more into data science in a corporate role or in an industry role. It’s a good idea. I’ve never thought of Upwork that way.
Nick: Yeah, you’ve got to validate what you’re learning in some way, right? If you can’t do that in your day-to-day job, you’ve got to find other avenues to do it. And there’s obviously a market out there, but how do you market yourself or your skillset for what you do? So, it’s interesting because a lot of the analysts and data people I’ve worked with in my jobs in eComm, typically they’re ostracized from the rest of the business because all they do is look at data.
But I think you’ll find they’re going to be the gold dust, they’re going to be the people that transform a business quite quickly, because they’re the ones that know how to actually bring people along the journey and what’s happening with their businesses. So if you can just distil that skill and take it and put it on something like Upwork, you know, something good could happen quite quickly.
Kirill: Yeah, for sure. So what are your thoughts on general data science? You mentioned that more and more businesses need to be data-driven and even the CRM skills or CRM work that you’re performing needs to be powered by data – otherwise it’s just like your opinion, or gut feel, on what you need to be doing. What are your thoughts on data science in general in the world? How is it developing from your perspective?
Nick: That’s a very broad question. I think people, especially people in an eComm role need to be quite aware of it because the way we make decisions now is not as linear as I think we all thought it was 20 years ago, for example. You know, “I want to buy something so I will go and buy it.” Now it’s, “I’ll look at 50 different sources on my phone and multiple devices.” There’s so much data happening with how you purchase now that it’s not as simple as just saying, “I have a product. Let’s buy it.”
Again, I’m talking in the context of eComm businesses, or people selling product, but anybody in any eComm role needs to really think about data and how that’s going to drive and what that’s actually telling you, because my role is all about who is going to buy next, when will I see them again. There is a science behind it, but it just depends on businesses and how they’ll get their staff to learn about potential data science techniques, or they just get a team in and they do it all for them.
It’s definitely a hot topic, especially in our arena. It’s a very hot topic, but I don’t think people understand what data science is and it’s just too mathematical. But if you look at, in the U.K., about what a customer is going to do, and propensity modelling, and attribution, and all of those types of concepts are very hot topics, and people always talk about it, but I don’t think they actually know much about it. There’s a huge opportunity there for people like yourself. It’s actually not as hard as everybody thinks, because there’s so many support tools out there to help make actionable insight from data. You don’t actually have to be a statistician or a mathematician to do this. It’s actually quite easy to do. In my subjective opinion, a lot of people just look at things like Excel and they’re like, “Yeah, that’s how you do data science.”
Kirill: Yeah. No, that’s not how you do data science most of the time. And that’s good. It means there’s opportunity to improve or opportunity to add more value just by using different tools or exploring ways to use those tools and suggesting them.
Nick: Yeah. Like, if you’re somebody just getting into the field—and again, I know I’m being very weighted towards eComm here, but if you’ve just come out of university, for example, or you want to get into an eComm role, typically they don’t ask for a lot of these skillsets. But I can guarantee that if you can create influence through actionable data insight in your business, it’s a massive skill that businesses will just eat up versus just saying, “Do you know how to use Microsoft Excel and Word and whatnot?” Somebody showing you a new dataset and you giving them insight from it quite quickly is going to be a very valuable skill moving forward.
Kirill: Yeah, I totally agree. And tell us more about eCommerce. You said that you use data science or data science techniques in the space of eCommerce. Can you give us an example of how you’ve taken eCommerce data and some modelling, or whatever you did with it, that helped get a good result in the end?
Nick: Yeah. I think the simplest way to explain this is you have a hundred customers, and typically eCommerce wants to know, “How many of those hundred customers will we see next year? If you purchased something, say in Black Friday, when will I see you again?” Understanding that data and the sort of trends of that, so if you want to cluster that—I don’t want to get too deep, but typically X%, maybe 20% or 30% of your customers, will return the next year. And you need to understand why that is: What is influencing them? What’s influencing such a small number of customers to come back? And that’s a great starting point because you can understand, “Well, how can I get more of those customers back? It’s going to be a lot cheaper to reactivate those customers than it is to find new ones.” But typically, in eCommerce roles, it’s all about finding as many customers as possible and typically they’re new customers.
You’ve got to take your customer on a journey. And what I’ve done, it’s pivotal that you understand why your customers don’t come back, how you get new customers, and essentially look at the dates in between. Do they come back a year on the dot, or do they come back every six months or do they come back every three months, for example? That type of insight. Or if you’re presenting to an array of stakeholders, quite quickly they understand what you’re talking about and you can start influencing what you’re going to do. Otherwise it’s just subjective thought.
Maybe we only get 20 customers back. Unless you show the data, nobody can really make actionable insights. CRM roles are quite cool like that, because they get to fuse the data with the insight and actually take action quite quickly. And then I use e-mail providers, so we have third party e-mail platforms—everybody would know MailChimp, but in eComm businesses you typically need a bit more power behind your database marketing. And a lot of these companies now have propensity models. I don’t know if you’ve heard of RFM – recency, frequency, monetary types of algorithms to sort of show where the clusters of customers are and how you can market to them quite quickly.
Kirill: Actually, I haven’t heard of that. Tell us more about RFM.
Nick: It’s a database mining technique, so you essentially you scorecard your customers based on the most frequent recency, so basically when you last saw them. So, you score them from 1 to 5 – it’s in quintiles, I think it is – and then you look at their frequency, so how often they come and purchase from you or visit you, and then your monetary is how much they spend with you. So, somebody that you’ve just seen that comes to your website — in my context, quite a lot — and spends a lot of money would have a score of 5/5/5. Somebody that I’ve seen a year ago came to the website once and spent 20 dollars, or pounds, would have a score of triple 1.
So what we can quickly do in e-mail platforms, or you can even do this yourself, you can market to all the triple 5 people and then you can market to all the triple 1 people and immediately use data to see—you can start testing and marketing to each of those scores to see where you could get a new customer, a returning customer back, or how much money you could make from them. It’s a bit of a hack definition there, but things like that are quite easy to do with some companies, because out of the box they provide this type of insight so you don’t have to do it yourself.
Kirill: That’s a really good walkthrough. So basically you might even market to them differently. If it’s a 5/5/5 customer, you might send them one message and a 1/1/1 customer, you might send them a different message because they’re less likely to respond, so it has to be maybe a more lucrative deal, or a quick sale or something, a cheap offer, whereas a 5/5/5, you know that they love your product and they’re more likely to buy something that’s high-end and you might want to approach them with that type of offer as well.
Nick: Yeah. I probably play more with database marketing, but it’s quite powerful. Depending on your resource, you only have so much breadth to do this stuff, but if you’ve got third party partners and they can make light work of your data, it’s very powerful because you can almost predict what your customer is going to do, or a percentage of what your customer is going to do, and you know when to market to them.
Kirill: Cool. And this is an interesting topic because I don’t think I’ve discussed this with anybody on the podcast before. We’re getting into customer segmentation. Let’s talk more about that. With your CRM background, you have a lot of experience in customer segmentation. What is geodemographic segmentation and why do businesses use it?
Nick: You’ve touched on one point. When people talk about customer segmentation, they do it geographically, they do it behaviourally. There’s many ways to cut the data and obviously every business is subjective, it has their own subjective goals. I wish there was always just this turnkey way that you could pick up the data and off you go. You sort of don’t know what you’re playing with until you start asking stakeholders the right question.
But typically, you apply the 80/20 rule, so “Where am I getting 80% of my customers from? Where am I getting 20% of my profitability from?” Because essentially you want to focus on them first. Otherwise you’re just spending too much resource.
Kirill: You mean 80% of your profitability?
Nick: Yeah. So, applying the 80/20 rule quickly allows you to take a lot of data to sort of get into the most actionable dataset because it’s just so much. Customer segmentation—typically I’ll look at geographical, so what countries we get the revenue from. For example, Europe—everybody gets so excited about Europe, you know, “Let’s market to German customers, Spanish customers, French customers.” In some cases they could be 10% of your customer base. And you typically have to translate everything you do in English into those languages as well. And then in countries like Germany, there is a lot of legislation as well on how you communicate to customers, so quite quickly you could burn a lot of resource going after the wrong customer. So you’ve got to think about your resource and what customer segments you’re actually going to go for.
Behavioural, we play a lot with when somebody comes to the website, what do we know about them. There’s a lot of technology out there which is quite cool where we can actually learn about whether you’re looking at the site on a mobile device or a desktop, and we can actually record your MAC address so we can actually start to get a 360 degree view of how you visit our site because, for example, somebody that looks at an e-mail will look at their e-mail on a mobile device, but then they might purchase on a desktop. That’s quite valuable information.
Kirill: Why is that valuable information?
Nick: Because if you know, for example, that somebody opens an e-mail on a Friday—there’s a cluster of customers that open an e-mail on a Friday and then they purchase on a Sunday night, you know that feeling just before you go to work on a Monday.
Kirill: You want to spend some money before going to work.
Nick: Yeah. It’s a weird thing, right? EBay, for example, one of their biggest traffic surges is that 7:00 P.M. to 10:00 P.M. part on a Sunday because everyone is like, “Oh, yeah, I’m relaxing, getting ready for the week.” So it’s quite valuable to see what cluster of your customers do something on a Friday on a mobile device and then actually purchase on a Sunday night through a desktop. You can actually make quite a lot of action on that. You could send them an e-mail on a Sunday afternoon saying, “Hey, just a reminder. There’s a great offer for you,” for example, and it hits their devices on a Sunday night just when they’re browsing just to get them over the line. Again, there’s a lot of ways you can play with customer segmentation, but the more people looking at it in your team will help. Sometimes it can just be yourself, so you’ve got to test, you’ve got to try this agile methodology where you go and test, get some insight, test it, see if anything happens, and then if it doesn’t, come back, do it again, find something else to test. Be quite systematic about it.
Kirill: Gotcha. And this is very interesting because it raises the question of ethics. Is it ethical to take advantage of people’s habits and send them on a Sunday night a reminder and get them to spend money? What are your thoughts on that?
Nick: It’s a grey area, especially in Europe. Obviously, when you go to a transactional website, you’ll see things like cookies. You see that box that comes up on the header or the footer saying, “We have cookies on our site. We want to learn a bit about you.” So a lot of what I’ve just spoken about is usually encapsulated within the cookie policy or the privacy policy. You might have heard about GDPR coming up as well. Have you heard about that?
Kirill: No. What’s GDPR?
Nick: Oh, God. (Laughs) It’s basically data protection. It’s a big legislation coming in Europe next April.
Kirill: Oh, yeah, I’ve heard about that.
Nick: Basically, business is going to have to declare all the data they capture.
Kirill: Yeah. It’s going to be such a pain for businesses.
Nick: Yeah. And it’s a big step because what you’ve just described is—you know, ethically, business is going to have to be very open about how they’re storing customer data, what they’re storing and how they transfer that data. I mean, it’s a good move. With what I do, it can get a bit Big Brother at times, but there’s nothing bad about it. But some customers who would’ve shopped before, 20 minutes later you’re on Facebook and you see this ad pop up on your feed saying, “Hey, Kirill! Buy this now!” and you’re like, “Hang on. I was just on this website. How does it know?” You know, things like that, that retargeting is a big part of eCommerce and it’s all encapsulated in the cookie privacy policy. And I think it’s just going to be more legislated as we go on because people don’t want to be followed all the time.
Kirill: Also, the way I normally answer that question is that it can be seen as evil, as unethical, but at the same time it can also be seen as helpful. It can be seen as, “We’re helping people not to forget to buy what they wanted to buy. We’re helping people to better see insights into their own preferences that they might not be aware of. But because we’ve seen many people like them who are interested in certain products which they’re interested in and they also bought this, that’s why we’re recommending this to them.”
And that kind of stems into what you mentioned, that people don’t like being followed around. But the question is, who doesn’t like being followed around? Is it baby boomers, our generation or the millennials? It’s a completely different story. If it’s the millennials, they don’t care. Most people of my brother’s age, who are 20 years or younger, they happily trade information about themselves for improved products and services. That’s the world they’ve grown up in and that’s the world they live in. So a lot of this legislation is still targeted towards our generation or older generations, but the new generation that’s coming, they’re not going to really care that much about it. That’s why, by the way, the cookie thing is going to be deprecated starting 2018, I think January or February 2018 it’s going to be removed. Websites won’t have to do that popup box anymore because it’s just annoying more than anything.
Nick: From memory, Google is going to have cookie-less ways of tracking you. I think Google already know, to be fair. I quite like Google Analytics and I use their premium package quite a bit as well. And the stuff you can do on that is just amazing, but it’s all legit because they basically hash you as a person. But under that hash code, they know your mobile device, your desktop. I mean, Google know a lot of it and absolutely you’re bang on. Organizations like Google will know quite quickly who you are even though you don’t think they know who you are. I can guarantee you they definitely know how you travel through the Internet and what sites you go on. I just don’t know if you can ever have a silver bullet approach to confirm to your customer that, “We’re actually just trying to send you meaningful prompts and not trying to stalk you.”
Kirill: Yeah, that’s an interesting one. And there’s a whole concept now of a digital avatar that as you browse through the Internet and you do stuff and you buy things, because these websites like Google and eBay or others, they collect information about you, slowly there’s an avatar that’s being built for you. Like in the movie “Avatar,” there’s like this creature. This is a digital avatar, a digital representation of yourself which is linked to your address, to your phone number, to everything that they know about you, plus all your preferences.
So basically, the more you interact with online stuff and eCommerce specifically, the more this avatar is enriched with information about you and at some point they could just recreate you – let’s say hypothetically – in artificial intelligence. And based on the behaviour of that avatar, they can predict your behaviour and therefore interact with you. Even before you have some desires or thoughts, they will know that you will have them literally a day in advance or something like that. That’s the extreme, but that’s where it’s all going.
Nick: It’s all of that now, right, but have you ever heard of the “Zero Moment of Truth?”
Kirill: I don’t think so. Tell us more about it.
Nick: Google commissioned this study, it was back in 2012. It was one of those turning points for me. It’s basically that moment, the point in the buying cycle when you’re researching a product. What gets you over the line? What makes you purchase that product? It’s that moment when your brain goes, “Yeah, I’m going to buy it.” They did this study, and I really recommend having a look at it because it’s really interesting and there’s quite a few videos on it. It’s what influences you to purchase. So you’re talking like this. It’s a lot like that. What behaviours mean that you will go and do something? How do you know when you’re going to buy a car? Is it lots and lots of research? And then you say to me, “Hey, Nick, buy that car,” and I go, “Yeah. You know what? I’m going to do it now because I’ve researched enough.” Why are you so influential in me doing that?
Maybe the zero moment, or ZMOT, is just not as relevant now because what you’ve just said is exactly—people just know so much they don’t need to think about it because they’ve got the data to support the behaviour so they can just augment that quite quickly and channel what they need to channel to the customer.
Kirill: Yeah, I know. It’s pretty crazy, this world we’re going into.
Nick: Slightly off piste, but yeah, it’s very interesting playing this data science concept because I definitely eat it up. But I won’t lie, if you don’t have that background, it’s quite hard to latch onto that stream of tech or skill that you want to learn to see whether you can actually make something of it. I’m definitely here coming from a very different end of the spectrum in the land of data science because everything that I’ve learned so far has always been far more of an analyst type of base. So we’ll see what happens.
Kirill: It’s interesting because the CRM space and the work that you’re doing dealing with behavioural aspects, that’s a huge part of data science. It can be attributed to analytics and the analysts and so on, but that behavioural aspect and psychology is like a goldmine. It’s always going to be required because people are always going to be there and you’re going to need to analyse it whether it’s in the sale, CRM-type of role, or it might be in medicine, it might be somewhere else where mass behaviour, mass psychology needs to be analysed through machine learning and other techniques in order to come up with actionable insights. It’s a great space to get into the field of data science.
Nick: Yeah, but it’s obviously so subjective to the business you work for and the client. And that’s why I really employ what you do because you tell people how to go on these journeys and I’m an example of that person that’s going, “Hey, I understand this. Now I want to go on this journey.” And it’d be great to do that in the CRM space, but it’s so subjective to the business that you work for because every customer is different. Every business segments their customers in a different way. But typically, they’ll be male or female and they’ll do some things.
Kirill: Age, location, you know, lots of things.
Nick: Yeah. I mean, a lot of the successes I’ve had is distilling all of this insight into like a one-page infographic on the new customer. Because people that don’t understand data, you show them something like an infographic and within seconds they understand what’s going on. It’s good to sort of pat yourself on the back and say, “Hey, people actually understand the data a lot more now through just these infographic icons.” So, it’s quite cool.
Kirill: Yeah, that’s a good approach. Do you have any more examples? That was a really cool Friday/Sunday example. I’ve got another one myself and then I’d like to hear if you maybe have some other segmentation or behavioural examples that maybe can be useful to someone working in the space of CRM. One that I know is that retail sales in February, they always drop off. They always have a dip in February. And that was so interesting to see, because in the Advanced Tableau course, I think we do some analysis of that and we look at the retail sector of Australia for all types of products in the retail sector, and they always drop off in every single sector, it’s crazy.
And the reason for that is because of the Christmas sales, because people buy a lot of stuff for Christmas for presents. But then you think, “Okay, so they bought stuff for Christmas in December.” You’d think it’d drop off in January, right? But no, it drops off in February because the credit card payoff days are usually 30-60 days or whatever, so it’s timed in such a way that in January, they still don’t feel the shock of the Christmas purchases that they’ve made, and then they get their credit card statement and then in February they’re like, “Okay, time to slow down the sales.” That’s always an interesting one.
Nick: Yeah. You’ve touched on what I would define as seasonality, or one of the many parallel streams of seasonality. And that’s such a great example. Another reason for that is people typically get paid in advance before Christmas. Then they don’t get paid again until potentially late January, so they can be on a shoestring budget and then they get their credit card bill and there’s this whopping payment where their January pay gets actually funnelled into their February credit card statement. That’s probably a very big factor as to why things drop off in February, but in the retail space, especially in the Western world – obviously I can’t speak too closely on the Eastern world, but I know in the Western world, things like—we see the same thing when we come up to Black Friday – if you know Black Friday…
Kirill: Yeah, end of November.
Nick: When I came over here, it wasn’t really a big thing in the U.K., and then about 5 years ago it just ramped up. People thought, “Yeah, we can make a lot of money here before Christmas kicks in.” So you actually have this like 8-week trading period where a lot of companies try and just funnel all this stock to their customers as quickly as they can to close off the year as profitably as possible. In the last couple of years in the various companies I’ve worked for, we do a lot of work in that September/October time to help get customers knowledgeable about our business. Because once Black Friday hits your inbox—I actually employ anybody listening to hop into their inbox and look at Black Friday and you will see that every single company you’ve interacted with probably sent you an e-mail on that day. You know, how do you get through that clutter?
And we spend a lot of time doing nurturing campaigns with our customers, so people that are on our e-mail database and they like to hear from us, we start to send them communication in the lead-up to big things like Black Friday to say, “Hey, there’s going to be a special deal for you. Come back here and we will send you this,” etc. But you’ve got to think about what your customer will probably get from all your competitors as you hit these key months.
And again, the way you do that is you look at all your sale spikes in a year and you start to sort of create these seasonality trends and you start to work towards campaigns to yield to those trends to hopefully see that they happen again. And if not, you just recalibrate what you’re going to do and still keep it going, but you need to have a bit of flex in what you do when you look at seasonality because it’s not always the same year for year, but it’s pretty close.
Like in the U.K., we see that drop-off in February as well because we’ve been marketing and advertising to people for eight weeks. You’ve got Black Friday, then you’ve got Christmas, then you’ve got an end of season sale. You’ve probably taken all the money you can. In a way, I’m contradicting myself. It’s not an exact science, but you can get pretty darn close to making sure you know what your customer is going to do in what month if you’ve got sort of clustering techniques on where you get your money from.
Kirill: Yeah, gotcha. And what size of databases have you worked with recently? Is it thousands, hundreds of thousands of customers?
Nick: Yeah, I’ve gone from half a million to about 2 million customers. In some cases that’s not very big – if you’re a bank, for example – but in a retail space it’s a good number to play with because you’ve got lots of different segments within that database. Typically, you’ll have a big portion that you don’t hear anything from, so that’s a great opportunity to reactivate them and say, “Hey, Kirill, remember us? Here’s 20 quid off! Come back.” Things like that are quite powerful.
Kirill: Yeah, gotcha. Okay – transactional data. That’s what I wanted to talk about. There are two types of data in geodemographic segmentation from my experience. There is static data, like what’s their gender, what’s their age – which changes, but slowly – what’s their income, what’s their affluence, how much money do they have. Generally those things are pretty static. Even if you record them every month, they’re not going to change that much over the course of a year.
But at the same time, there’s also transactional data. There’s data that they swipe their credit card—if you’re a bank, you have them swiping their credit card. But if you’re not a bank, you’re a store, you can track what have they purchased. And not only what they purchased, but which pages on your website have they been to. Like you said, have they used their mobile or their desktop or laptop or whatever, what browser did they use, what time of day did they do it, so data that is updated frequently, like on a daily basis or even more frequently.
From your experience, the difference between the two, the value that each one of those types of data brings into your analytics, into your segmentations?
Nick: The most valuable part is the way customers pay, so do they pay with Visa, MasterCard, American Express or PayPal, for example? In Europe, it’s massive that you understand this because in Europe there’s different payment methods that people prefer that aren’t typical in the U.K. So if you want to go to Germany, for example, you need to employ a different payment gateway because a German customer likes to transact in a particular way. If you don’t know that, you are missing out.
My experience, I look at the payment method more within transactional data. And one of the things that you do in the fashion space, you actually look at the titles, so Mr and Mrs and whatnot, because you can find some people—males use their other halves’ credit cards, and they actually use each other’s credit card but the delivery address is different from the cardholder address. So some people pay with their other halves’ card at times.
Kirill: What action do you take on that? How is that insight useful?
Nick: Logical question. So what?
Kirill: Yeah, exactly.
Nick: That’s the question you always need to ask yourself. That’s interesting, so what? I would say it’s probably more of a just nice to know, that if our returning customer, our loyal customer, elicits that behaviour, it proves typically, potentially what is maybe a geographical trend that we see, so maybe in a more affluent part, that behaviour is more prominent than in a less affluent part of the country. I wouldn’t say it’s too actionable. It just helps support who is the customer and then how do they transact.
Things like PayPal are big kickers because you have to put a bit more development into your website for things like PayPal to make it a seamless experience because you have to hop off the website, go to PayPal, and come back to the website. But it’s such an easy way of transacting because you’ve just one login and off you go. You don’t have to type in any credit card data. And we see a rising increase in things like PayPal, so you need to think about, “Okay, how can we make the customer checkout experience a lot more fluid using things like PayPal?”
So they’re the two areas that I’ve played with in transactional data. Other than that—eCommerce is a global thing, so you want to look at where you are getting all your transactions from and sometimes you have to put currency buffers in and does that make you more expensive, for example. That’s probably an arena that I’ve played with in transactional data.
Kirill: Interesting. And I’ll probably add a bit to that. When I was in Sunsuper, the pension fund in Australia, it was interesting because we had access to a lot of—like, any pension fund, not specifically Sunsuper—the difference between a superannuation fund, or a pension fund, and a bank is that a superannuation fund has access to static data mostly, you know, gender and location and affluence, whereas a bank has access to all that data plus transactional data. They have the advantage.
In my view, it’s a huge advantage that you have information on what your customers are purchasing all the time. You can build a much more accurate avatar for your customers. I was working with a senior data science consultant from overseas; he came and was helping out with some projects. He saw that we don’t have the transactional data and it was like a revelation for him. Without transactional data, the accuracy of the models drops by 30%, something ridiculous. When you have transactional data that’s constantly changing, that’s constantly being updated and you have information on a daily basis about people’s purchases, preferences and so on, you can understand what type of person they are, segment and cluster your customer base much more accurately, about 30% more accurately than when you don’t have transactional data. Yeah, that was a very interesting insight for me.
Nick: You know, one of the things I did forget to mention is we looked a lot about what our first and new customer purchased. Typically, they would purchase a particular shoe type, for example, versus returning customers. And it’s quite interesting because then you can funnel all your marketing efforts to actually promote that one product. Because typically you’ve got the data, the transactional data to prove that is one product that’s preferred by first-time customers.
And it’s the same with any business, I’m sure. Whether you’re selling products or services, there will always be one to three products that are purchased more than others. You need to make sure that they are just watertight and they’re fantastic so you always have that steady flow of customers purchasing them.
Kirill: That’s again your 80/20 rule, right?
Nick: Yeah. I mean, you’re directed through and through with eComm teams, but typically you’re wearing many hats. You’re not just brought in as a subject matter expert for what you do, whether it’s SEO or paid search or socially. You have to be quite commercial and you have to understand the full customer journey and how you influence that with what you do.
This is why employing data science methodology into how you work is just so advantageous, because you’re not just doing one thing each day. You’re trading a business and you need to be quite fluid in your influence. In my subjective opinion, you really need to understand what the data is telling you and how you can make insight on it.
Kirill: Yeah, totally. I have another question for you. A/B testing – do you do much of that in eCommerce?
Nick: Yeah, a lot. Again, it depends on how many humans you’ve got in the team to help you do it because it takes time. Typically, what we play with each week is doing things like e-mail subject lines, because that’s the first thing that you see in an e-mail. A big thing everybody loves is emojis. I’m a bit 50/50 on them, but again, do that Black Friday thing, have a look at Black Friday, and I guarantee you, you will see so many emojis in your inbox. We A/B test that, but then we also A/B test how people come to the website. So if we know that Kirill is a very profitable customer—
Kirill: Which Kirill is not. Kirill doesn’t buy a lot of stuff. (Laughs)
Nick: Yeah. You told me that. (Laughs) If you come to the site, I might show you a particular message versus a new customer that I’ve never seen before and just test that engagement. And maybe you’ve come to the website five times, which shows that you’re ready to purchase, or you’re considering purchasing, so maybe I could have a popup that comes up, or some type of activation to prompt you to say, “Hey, if you check out now, here’s some incentive. Or why don’t you have a look at this?” There’s a lot of technology that you can do that’s quite easy to A/B test how your customers, whether they’re transacting customers or just browsing customers, interact with your site.
But again, you need to think about it. You need to think about the customer journey and what you’re actually trying to prove. Because typically with A/B testing, people can get quite excited and you could have like a hundred tests. But if you don’t have a big team, you don’t have that cadence to do those hundred tests, so you have to be quite picky about what you test.
Kirill: Interesting. Can you tell us the technicalities of the A/B test? If somebody wants to run an A/B test, let’s say they have a small team, a small customer base, what are some of the steps that they need to take in an A/B test and what are the most important things to look out for, like some pitfalls you might help people avoid?
Nick: Yeah, typically you’re trying to define what the customer is trying to do, what are you helping them solve. And that’s largely how I’ve always done A/B testing. You’ve got think about what are they doing now and what do we think we want them to do based on data. What do we think we should influence them in doing? Depending on the stakeholder group, you need to think about how people can sometimes influence the wrong type of test, so you need to be quite specific about what you’re trying to test. And you’re not testing an array of things. You’re testing, “do our customers resonate more with a blue button versus a green button?” That’s essentially a test. It’s not lots of different messaging so people can get pretty hyped up.
And some of the pitfalls I’ve seen is, if you get the wrong stakeholder group in there, they can say, “Let’s just test everything,” and it’s all very broad and it’s not specific enough and people aren’t thinking about what you’re actually trying to help the customer do. And there’s a big thing around business goals and customer goals.
People typically always talk about business goals, “We want to make money.” But customer goals are “I’m a customer. I’m on your website for a reason. How are you going to help me do that?” So if you start thinking like that, it’s quite a challenging exercise to think about business goals and customer goals, but it helps things like A/B testing quite quickly. And it’s called—I’ve worked with a UX designer, so it’s sort of a little bit of the UX part of the discussion now, and that’s user experience. And I’ve worked with quite a few consultants and they talk like this because they need businesses to understand what their customers are doing and what we’re trying to do through A/B testing.
And it’s a very interesting journey, but you’ve got to think about what the customers are doing and what you want them to do, but where are the pitfalls? So typically, a lot of A/B testing that we play with is in the conversion part, getting people to transact. There’s usually some blocker there. Kirill, maybe you come to the website and you want to transact, but then I present you with this very long form of about 30 different fields. Typically, as a new customer, you’re not going to fill out all of this information because you’ve never transacted with me before, so maybe all I need to give you is just “Tell me your e-mail address and your payment details and off you go.”
That’s quite a good test to play with, but it’s quite a massive one because it’s part of the checkout. At the end of the day, it all rolls up into what do you want the customer to do, because otherwise people in your business will just say, “Let’s test this.” You know, why do you want to test that? What are we helping the customer achieve by testing that?
Kirill: Yeah, that’s cool. It was very interesting, what you mentioned about business goals versus customer goals. Do you have an example of when you have, or you would have, or a company should, forego a business goal or sacrifice a business goal in order to help a customer instead?
Nick: Ooh, controversial question.
Kirill: I know, yeah.
Nick: Controversial question! Oh, God, I don’t have enough experience yet in this, because I’ll always say business goals supersede customer goals, which is horrible for me to say but it’s very hard for eCommerce businesses to drop the speed that they make money to yield towards customer goals for longer term gain. In eCommerce it’s very hard. I think in brick and mortar sort of offline businesses it might be a bit easier because it takes a bit longer, but eCommerce rolls can be quite—you know, you’ve got these budgets you’ve got to hit every week so it can be quite…
An example of what we did is we come into the summer season, right, and I want to buy a pair—well, in Australia we call them thongs, of course. (Laughs) In the States we call them flip-flops. Essentially, as you come in, sort of 2-3 months before a holiday period, people start coming to footwear sites to have a look at what seasonal footwear there is. So, a customer goal is, “I need a pair of shoes for my holiday that is fashionable and open-toed.” But a business goal is, “We have to make this amount of money in this week.” They’re very different goals, right? So your question is very valid, but it’s very hard to not weight everything towards the business goal but you need to be very mindful of the customer goal because a customer goal, if you fulfil that, what you would actually do is create a lot of content to help that customer fulfil that journey.
So you tell them about the shoes, you show them what new shoes are there, how it can benefit them, future benefit. And then you would hope that the business goal is fulfilled because you’ve done that work to take them along the journey to then purchase what is their [indecipherable]. So the two eventually connect, but depending on the eCommerce business and maturity of that business and how customer-centric they are, I’ve always seen it to weight a bit more towards business goals. But get a UX consultant into your business and they can actually help you transform quite quickly the way you think about a customer. Because that’s their job, it’s to make the customer experience a lot more fluid and beneficial for your business.
Kirill: Have you heard of this company in the U.S.—it’s a shoe company so you might have heard of it—they really made a huge name for themselves for the customer experience that they were creating. People would call them up and there was even one case when the support rep was on the phone with a customer for like 10 hours in a row helping them pick out the right shoes. It’s a big company—
Nick: Zappos, right?
Kirill: Zappos, yeah. That’s one example that comes to mind of when they put the customer experience ahead of the business goal. It’s no good for any company to have a customer support rep sitting on the phone with one customer for 10 hours. That’s a lot of cost. But they do that and they built a name for themselves and now people know that if you go to Zappos, they’ll guarantee they’ll find the right size, find the right shoe, find the shoe that they want. They sacrificed a bit of business revenue and so on to build this brand, to build the customer experience and now everybody refers them to their friends, to Zappos.
Nick: Yeah. If you’re ever in Vegas, you can do a tour of the Zappos’ business. It takes about 2 hours or so.
Kirill: You did that?
Nick: Yeah, and it was amazing. I was there with my wife and she was like, “You need to calm down. You’re a bit excited.”
Kirill: Because it’s a shoe business, right?
Nick: Well, actually it’s much bigger, but their methodology—I mean, in the U.K. we can be a bit cynical at times, but the American way of customer service is just so fantastic. You’ve got these commandments and all their desks are themed. And then you hear about that story that you just spoke about and you’re just like, “People like me just eat it up because I’m like this is fantastic. These guys are so customer-centric that every single thought starts from the customer and then they work up. You know, in many other businesses, if you’re small it’s about “Money, money, money – let’s generate it,” but at Zappos their culture is completely fuelled by the customer. And all businesses are, but these guys brag about it, you know, there’s books about it.
And it’s such a cool company to go to because you want to come back from a tour like that and say, “Hey, let’s employ some of these methodologies in our business.” But then you’ve got a huge cultural journey then. You’re basically asking the business to yield to how you’re thinking about the customer. I think CRM roles should be very closely aligned to UX designers and data analysts because the three of you, the three departments, can actually start influencing a business to be far more customer-centric and think like Zappos as well.
You know, they are extreme cases, but it’s interesting that you talk like that because there’s not many companies that can talk like this where they’ve had staff on the phone for hours on end. I think they get customers just calling in for a chat, but if that customer comes back and buys maybe four or five things from them, you know, expand that out by another 100,000 customers, there you go. A new customer base. I mean, you’ve got to figure out what works, but it’s a—yeah, if you’re in Vegas, go have a look. It’s really interesting. They actually have full tour guides taking you through the business. That’s quite cool.
Kirill: You’re a very interesting guy, Nick. Out of all the things you could do in Vegas, you go for a Zappos facility tour.
Nick: Yeah, yeah. Others go to Vegas to gamble and I go for a tour.
Kirill: Fair enough. Better than going and gambling your money away.
Nick: I did go gambling just for [indecipherable]. I probably should have said that first.
Kirill: Of course you did. And then when you lost all your money, you were so devastated you were like, “I’ve got to go have some fun on the Zappos tour.”
Nick: Yeah, “I’ll just spend 20 bucks and go on a Zappos tour and drown my sorrows and walk around a building for 2 hours.”
Kirill: Yeah, gotcha. All right, Nick, we’re coming to an end. It’s been a pretty cool session. I just have one question left for you: Do you have a book in mind that you can recommend to our listeners for them to get better at CRM stuff, data science or something that can help better themselves?
Nick: Yeah. I haven’t found the Holy Grail book, but the one that I refer to the most is Cole—I can never pronounce her last name, but she—I know a lot of people that you’ve spoken to mention her, Cole’s “Storytelling with Data” book. I use that a lot because it’s a great way to distil how you think you’re analysing data and actually how you should present it. So I use that book a lot as a reference tool. I’m going to put one more in. There’s one called “Resonate” by Nancy—what’s her name again? Yeah, Duarte, I think it’s Nancy Duarte. There’s people that do these great storytelling books with data, but Cole’s is probably what I use the most. I’d highly recommend anybody getting that. If you’re in this world where you need to distil a lot of data insight into an infographic or a presentation quickly, it’s a bit of a Holy Grail.
Kirill: That’s pretty cool. I’ve actually got a book, I think, myself—one of the guests on the podcast recommended it and I thought you might be interested in it because you’re interested in visualization. It’s called “Dear Data.”
Nick: Never heard of it.
Kirill: It’s like these two ladies, one in the U.K. and one I think in the U.S., for a year they were sending each other postcards but instead of writing text in the postcards, in each postcard they would describe their week with charts. So check that out. That’s what they did in that week, what the other one did in that week. They were like hand-drawn charts on these postcards, so here you have all of these 52 charts for each one of them.
It’s very interesting. It gives you some cool ideas for data visualization and stuff. You know, you just open it up and you’re reading somebody’s postcard, like once an evening you check out a postcard and so on. I haven’t read many myself, but I read a couple and they’re interesting. For instance, they might describe an hour clock and each hour they have something—like, a little symbol represents what they were feeling, what they were doing. It’s very cool.
Nick: It’s the hardest skill, but the most fulfilling when you can tell a story with data. As you probably can tell, I can talk. (Laughs) Sometimes you just need the visualization to do the talking for you. That’s why I’m putting a lot more time into it, because in one room you can bring a CFO, a cleaner, and your boss and tell them the same story quite quickly, so it’s quite cool.
Kirill: Yeah, gotcha. All right, thanks a lot, Nick, for coming on the show. Where can our listeners contact you or get in touch and find out more about your career and journey? What’s the best way?
Nick: Hop onto LinkedIn, Nick Pape on LinkedIn and you’ll see my profile there. I’ve got a presentation on CRM that I’ve put up there as well.
Kirill: Which has 20 views. (Laughs) By the way, easy way to remember: Nick Pape is like Nick Paper, but without the R on the end. That’s how you taught me. I can’t get it out of my head, man. It’s like just Pape.
Nick: You’re not supposed to remember that part, mate. You know, I’ve done that and to you it’s fine, right? I’m never going to use that technique again because people laugh too much.
Kirill: Well, now maybe a hundred people, maybe a thousand people are going to hear it.
Nick: Several hundred thousand, yeah.
Kirill: And you’re going to be in their heads as well.
Nick: “Oh, so you’re Paper without the R.”
Kirill: Yeah, yeah, exactly. All right, thanks a lot for coming on. This has been a pleasure and a great chat today.
Nick: Thanks a lot. Cheers!
Kirill: All right, there you have it. That was Nick Pape, digital eCommerce specialist. I hope you enjoyed today’s episode and learned quite a few things. My personal favourite from this episode was probably the whole concept of how Nick is learning data science to maybe move into data science, but also maybe stay in eCommerce, but borrow those most powerful concepts that he’ll take. I think it’s a very interesting strategy. Maybe there are people listening to this podcast who are happy with their roles, who are happy doing what they’re doing and don’t necessarily want to become data scientists, but they do recognize the power of the data. If you are one of those people, then like Nick, you can just learn data science to borrow those most powerful concepts to enhance your career. Very interesting strategy, and I totally see why that could be the case.
And make sure to connect with Nick, you can find the URL to his LinkedIn profile and all the resources for this episode at www.www.superdatascience.com/97. And by the way, of course the video for this session is there as well. And on that note, thank you so much for being here today. I look forward to seeing you next time. Until then, happy analysing.
