Charlie Cole, global chief eCommerce officer at Samsonite and chief digital officer at Tumi, emphasizes the importance of analyzing data within your own company while utilizing an unbiased third-party.
In-house staff analyzing data seems obvious, but an unbiased third-party can be beneficial Charlie Cole— global chief eCommerce officer at Samsonite and chief digital officer at Tumi - tells TechRepublic's Tonya Hall. The following is an edited transcript of the interview.
Tonya Hall: Machine learning, artificial intelligence, analytics, China, Korea, Japan, and a leather and camo backpack. Welcome, Charlie.
Charlie Cole: Thanks, Tonya.
Hall: That's a pretty long title. Tell me, what's the story behind that?
Cole: Yeah, it's sort of an amalgamation of two, because I was the Chief Digital Officer at Tumi. Tumi was acquired by Samsonite, and a lot of people, including myself at the time, don't know this, but Samsonite is kind of a holding company similar to that of Procter & Gamble from a structural perspective, and so Samsonite owns Samsonite, Tumi, eBags, High Sierra, Gregory, American Tourister, Hartmann, Speck, and some other brands around the world. Soon after the acquisition, I pivoted from my role as just Chief Digital Officer of Tumi, but also oversee the global portfolio digitally at this point, so maintain my relationships directly with the Tumi team, but now also work with the portfolio around the world.
Hall: About, that's huge. Right? You've got a lot of different brands that fall underneath your repertoire. What are some of those?
SEE: IT leader's guide to the future of artificial intelligence (Tech Pro Research)
Cole: I mentioned some quickly. Probably the biggest ones that people recognize is, obviously, the Samsonite brand. eBags.com is an acquisition we made last year. We own Speck. Speck is a cell-phone case-brand based out on the West Coast. We also own Hartmann, which is a kind of a legacy brand that a lot of people know on the higher-end space, and then we also work with High Sierra, which High Sierra is a backpack brand that I feel like a lot of people have heard of. There are a set of eight main brands, and then we do have some regional brands in Korea and in Latin America as well.
Hall: You manage digital strategy, and that includes social sites and things like Facebook and web traffic. How do you pivot when the digital game changes, like when Facebook changes their algorithm, for example?
Cole: Well, I think there's a lot of stuff that we can control, and there's a lot of stuff that we can't. I think one of the coolest things that's happening in the brand space, and I'm going to go beyond just retail, but brands in general, is, there's sort of this tug of war between whose job is it to react to one another? You know what I mean? I'm kind of, in a tertiary way, referencing the data-protection problems that Facebook has, but I think our job is really to respond to the algorithms as opposed to dictate the algorithms.
We are constantly analyzing a myriad of factors using on-site analytics, using store traffic to try to get an understanding of what that algorithm is, because depending on the advertising platform, like a Facebook or a Google, the changes can be pretty clandestine, or the changes can be out in the open, and so I feel like our job is to analyze our first-party data at predominantly the analytics level, and so whatever your platform may be, Google Analytics, Omniture, Webtrends, whatever, Coremetrics — so I'm dating myself — that's kind of where we do the majority of our analysis, to try to decipher what really matters to Facebook.
We obviously have the metrics that come from Facebook, but there's a bit of a conflict of interest of Facebook or Google, or any ad platform, to tell me something that doesn't help their bottom line. I think that understanding where algorithms may change, yeah, you're going to get analytics, and yes, you need to analyze your quality score, but also how the traffic responds on your site is a piece to understand how traffic is performing on the platform, because that is the conflict of interest that I feel like advertisers maybe are not jaded enough about is the fact is, Google and Facebook and Amazon with AMS, these are not benevolent meritocracies. Right? These are businesses, and they're trying to make as much money as possible off of us, and off of every other advertiser on there, so trying to manipulate their algorithms for our gains is going to be in the completely opposite direction of what they're trying to accomplish with algorithm changes. I think that's why you have to look beyond just the numbers Facebook gives you and try to analyze it on your end as well.
SEE: Hardware spotlight: The Raspberry Pi (Tech Pro Research)
Hall: Do you analyze those metrics internally, or do you use third-party vendors for those?
Cole: Great question, and the answer is yes. Right? I am a firm believer in always having an external analytics partner as well. Internal analytics are, I think, sacrosanct. I think as an eCommerce person, everybody on our team has to be able to analyze their own business. Okay? Totally, you need an internal competency. I, even now, try to do one dedicated analysis a day, right, where I just myself am diving into the numbers, aggregating the numbers. Right now, I was doing an analysis trying to make a correlation between impressions on Facebook, AMS, and Google, non-brand, branded, PLAs, mobile versus desktop, and correlate that against our in-store traffic. Right? That's a bunch of stuff. The reason why I complement my own analysis and our teams' analysis with an external team is, brands can really talk themselves into anything. Right?
You are, it's almost impossible to get around your own biases of living in a brand. I remember one of the first quotes that I remember from my career came from this guy named Tim Ash. Tim Ash was talking specifically about on-site usability, and he said the quote, "No one likes to be told their baby's ugly." I think it's really true, which is why, as a third party, like if you're a dermatologist, and you go to pay that dermatologist to go see them, and they're like, "Dude, your skin's pretty messed up," that's going to be received much better than if your brother says it. I think that's why I always like to complement our in-house capabilities with an external analytics team, and we work with a team that, frankly, doesn't care about our brand and thinks of us just as a client.
Hall: How do you go about selecting external analytic partners?
Cole: Well, I think ... There's, obviously, capability things. My single largest factor in any vendor selection, and I don't mean to over simplify vendors, but any one, is sort of backdoor references, if you will. You could hand, provide me references, and of course they're going to be like, "Oh, my gosh, Tonya is the most amazing thing on earth," but I'd rather go to your LinkedIn, find people I know that you also know and get, if you will, the real story, where ... You've taken a reference for work. I've given references for work. You guys obviously talk about it. You say, "Hey, this is what they're really looking for. Make sure to play up my ability to host podcasts," whatever it may be, but the truest form for me is getting these friend-to-friend, peer-to-peer references, and not the ones hand selected by the vendors.
An analytics provider, it's a really hard thing, Tonya, for me to talk to somebody on the phone and be like, "Oh, yeah, that girl's better at math than that guy." Right? I mean, I can try, and I was pretty ... I was a statistics guy myself in college, and so I like to think I can decipher that. The reality is, money talks, and so if I can find out their client list and then talk to the references they didn't give me that are in my peer group, that is the most powerful thing to me, for as far as selecting a vendor for the first time.
Hall: You've been collecting customer and market data for decades now. How has machine learning and artificial intelligence helped you make sense of some of that?
Cole: I have this saying, which anybody who's listened to me before has probably heard me say, which is, we try to stop guessing. Right? What machine learning allows you to do is have a very brand-agnostic view of a very specific question. You and me can sit here right now and say, "Here's our problem. Tonya and Charlie are going to figure out who are the best 100,000 people to mail about this camouflage backpack." Right? You and I are going to go, "Well, we want to look at the people who have viewed camouflage backpacks. We want to look at people who have bought them in the past, and we want to look at people that have opened camouflage backpacks. How many people is that? Around 100,000." That's a great list. That's smart, right, and we had a good hypothesis, and we did a good job.
The reality is, machine learning can do that exercise a million times before we can and incorporate nth degree more data to get a more accurate, representative sample. Maybe what we didn't have time to munch is that there is a correlation between people who have bought leather in the past and moving to camouflage. We don't know. You and I could never come up with a hypothesis that incorporates that and finally arrive at an answer. Machine learning can do all of those questions and then come up with a correlation that's far better than what we can. It allows you to do data queries that, candidly, you would never think of. That's not to call me, or you, or the collective us stupid. It's just impossible for us to have every single iteration to be effective.
SEE: Research: Companies lack skills to implement and support AI and machine learning (Tech Pro Research)
Hall: Okay, I think you just outed me for wanting a camo backpack, but okay.
Cole: It was not a random product selection, I'll be the first to admit it.
Hall: I'd actually like to request camo with leather, but anyway.
Cole: Yeah, and see that? That's how good my machine learning is ... It went through my auditory system and found that you were a leather correlation, which is why I used that example.
Hall: Let's talk about-
Cole: It's total BS.
Hall: Let's talk about the eCommerce elephant in the room, right? Let's talk about Amazon and Alibaba. What do they mean for luxury vs. commodity brands? I mean, how do you maintain your brand's prestige, like Tumi, for example, when selling through mass-market websites?
Cole: Product segmentation's a big deal. Enforcing some sort of pricing in the market, MAP pricing is or is not legal, depending on what region you're in. Not an option in Europe, but in the U.S., we can enforce it to a certain extent. I think pricing's a really big deal, but to answer your question really simply, I don't really know how a commodity brand survives if they're not surviving on Amazon and Alibaba, and in Latin America, MercadoLibre. I just don't know how, and I've ... I don't know how the Calvin white T-shirt, Calvin Klein white T-shirt survives. Right? I just don't know how that happens.
Now, for Tumi, we have to do a couple of different things. We have to make sure we're aware of the fact that the majority of the people in both the United States, and China, and probably in Latin America, and probably in Europe, are starting their product searches on one of these marketplaces, whether it's Amazon, Alibaba, JD, MercadoLibre, Zalando, whatever. Just make your peace with that and then decide, "Do I need to be there to at least let those eyeballs know I exist?" For Tumi, what we do is, we give a subset of what we call our core product to these marketplaces, the things that we think are the best distillation of Tumi as a brand, and also what that customer is looking for.
From there, they can discover our brand, buy our brand, but we hold back certain products, and services, and offers specifically for our direct-to-consumer channels. There's no silver bullet, Tonya, but I think it starts with product segmentation as a luxury brand, or if you decide not to be there at all, your full-time job is to combat resellers in the program. If you're not selling on these marketplaces, no one can, and you have to do everything in your power, from a distribution perspective, from an agreements perspective with your partners, because otherwise, the floodgates are open. Right?
If someone can find your brand in a marketplace, nine times out of 10, I would argue, an Amazon buyer has no idea where they're buying from, and they don't particularly care, as long as the user ratings are really good, and so you either have to be there and do something to make your brand stand out, but still represent your brand, but keep something back for direct to consumer, or don't be there at all and make sure everybody else can't be there either with your brand, which, I would argue, is even more challenging.
SEE: Artificial intelligence: Trends, obstacles, and potential wins (Tech Pro Research)
Hall: What are some of the potential antitrust concerns for Amazon, Google, and Facebook, for that matter?
Cole: Well, you ... I'm not a lawyer, but you look back at how people have getting ... People got whacked in the past. Right? Then one that jumps to mind, for me, is the EU case where Google was giving, from their perspective, giving preferential treatment to their own product and their own advertisement results. Now, Amazon has a lot, a lot of private label brands, and they're making more and more money on AMS. What happens when you land on a homepage for Amazon, and the only thing you see is Alexa, and the only thing you see is AmazonBasics product, and then you go and you type in "luggage," and the sponsored products, the only thing you see is AmazonBasics?
To me, that is the exact same thing that happened in the European antitrust case, which I think the settlement was in the $2.4 billion range. I just don't see a scenario where there is not at least an analogous violation, and I wonder how sustainable it is. It's all smoke, no fire right now, but I think that it's getting dangerously close, and when you look at the acquisitions going on, what happens if you type in "ranch dressing" into Amazon, and everything that shows up is Whole Foods 365? That, to me, is ... There has to be at least someone that points to that and be like, "Hey, isn't it like that thing that happened in the EU? Because it seems to me like the exact same problem."
Hall: Digital transforming the travel industry, you've been working on that for a while, and in fact, you've got a long history of other brands, other legacy brands that you've been transforming. What is the biggest challenge for digital transformation and travel?
Cole: Well, you actually kind of answered the question in the question, which is, we're traveling. Right? The best customer we have is going to view our stuff, like content, websites, emails, or whatever you want to call it, in the United States, Japan, Europe, Australia, and Russia, and China. Those are inherently different data governance challenges. Right? Think about my life before GDPR and my life after GDPR, and how can I follow a U.S. consumer when they were opening emails around the world? Well, heck, I can look at their open-send-click behavior and store it wherever I want, even if they open it in Paris. Nope, not true anymore. What happens if they're behind the firewall in China? Nope, not true anymore.
SEE: Enterprise IoT research: Uses, strategy, and security (Tech Pro Research)
Keep in mind, this is my best customer. This is the person traveling 200,000 miles a year, and I have to come up with some sort, or I shouldn't use "I," we have to come up with some solution that allows us to at least have enough data to support this person. If their bag breaks in China, and they go to a store in China to get it fixed, it doesn't matter to that customer if they bought it on tumi.com in the United States. It just does not matter, but it comes with a remarkable data governance challenge, allow us to effectively service our customer. Because we're travel, all the things you hear about governance around the world are a reality for our very best customers, and it creates an inherent challenge that we try to have a lot of fun solving.
Hall: You're an analytics guy, so what are the biggest challenges to analytics today? I mean, are deep learning, artificial intelligence the silver bullets?
Cole: They help. Right? They help, but I would argue if your name isn't Facebook, Google, Amazon, Netflix, the biggest challenge to analytics today is that you have a really hard time tracking your user across their multitude of devices. We were joking before this, Tonya, that I was in the process of silencing my Google Pixel 2, my iPad, and my iPhone 10, and I'm talking to you on my work computer, so with most analytics platforms, I'm four people. Right? I'm talking to you on four different devices. Now, if you're Google, Facebook, Netflix, you have this stream. You have a login that allows to say, "Hey, here's Charlie on his phone. Here's Charlie on his iPad. Here's Charlie on his computer," but that's not something they're going to volunteer to you.
Go back to the first thing I said. Google and Facebook is not financially motivated to help you solve this problem, because you will become more efficient with your advertising, and you will spend less money, so as a brand and as a non, call it, data gatekeeper, which those kind of four people really are, that's an inherent challenge for us that we have to do a combination of relying on the data given to us, which, again, is scary and a conflict of interest, and do our best to stitch it together using some thread which more often than not in the past has been a cookie or an email, but those are sort of under siege as well right now.
Hall: In the past, people would maybe search for products on their phone but rarely buy them. They'd actually move to their desktop or laptop and actually complete the purchase. Have those trends changed, or-
Cole: Yeah. It's just a matter of the orders of magnitude. Like in China, I mean, forget about it. Desktops are almost dead. A lot of people believe cash won't exist in China in 2030, so I just think that that orders of magnitude that happen in China are out of control. Korea, Japan, not to that extent, but yes, it's now more people are buying on mobile phones than on desktops. In the U.S., the trend has pivoted, but I would argue that I would bet for the majority of brands ... I'm not talking about multi-brand retailers. I'm not talking about Amazon. The majority of brands, the Nikes, the Tumis, the Calvin Kleins, the majority of their traffic is mobile, but the majority of their sales is still on desktop.
There's an interesting meta there, Tonya, which is very interesting. For some reason, for all of our technological advancements, U.S. has been really slow to adopt cashless. Really, really slow. Like paying with your phone, I bet you have one of the most technical audiences on the planet, and I would bet the ratio of people that have paid for things with Apple Pay or Android Pay is probably maybe above 50%, where that number is in the high 90s in China. I can't really explain it, but I think that's ultimately holding back that mobile conversion rate and ultimately mobile being the majority of the revenue.
Hall: Well, Charlie, I purchase on mobile, and I'm going to be purchasing one of-
Cole: So do I.
Hall: ... those camo backpacks, I think. Thank you so much for your time. If somebody wants to connect with you, maybe they want to find out more about your work or maybe they want to just connect with you personally, how can they do that?
Cole: I do, I am an active LinkedIn user. I know I might be a dying breed there, but please do follow me on LinkedIn, and then I'm an old-fashioned believer in one-to-one communication, so email me. My email is charlie.cole, C-O-L-E, @tumi.com. You can email me anytime.
Hall: All right. Thank you so much, Charlie. I really appreciate your time, and you know what? You can follow me by going to TechRepublic, or maybe go to my website, which is tonyahall.net. I've got links to all my social platforms. Facebook, LinkedIn. In fact, if you'd like to chat, find me on Twitter. I'm on @TonyaHallRadio. I'd love to hear from you. Thanks for watching.
- How to implement AI and machine learning (ZDNet special report) | Download the report as a PDF (TechRepublic)
- How the GPU became the heart of AI and machine learning (ZDNet)
- Machine learning: A cheat sheet (TechRepublic)
- DeepLocker: When malware turns artificial intelligence into a weapon (ZDNet)
- Report: The 10 most innovative companies in machine learning (TechRepublic)