Podcast: Brian Eberman Joins Gen Furukawa To Discuss How To Predict Customer Value And Use This Data To Optimize Your Marketing Campaigns
Zeenk’s CEO, Brian Eberman, joins host, Gen Furukawa, on his podcast “Cart Overflow” to discuss how e-commerce brands and marketers can pull actionable insights from their data to better optimize their campaigns and increase profits.
Included in this discussion:
- Looking at customer value versus CPA metrics
- How to use analytics to predict future value and churn probability
- Using CLV vs CAC ratios to optimize your spend
- Understanding your sales channels, ex- DTC and Amazon, and how they work together
- The state of ad measurement and tracking
0:00:00.6 Gen Furukawa: Alright. Hey, everyone, welcome back to another episode of Cart Overflow. I’m your host today, Gen Furukawa, and we talk about e-commerce marketing strategies, tactics, how brands are driving growth. A lot of that is contingent on the analytics that drive those decisions, and today we have Brian Eberman who is the founder of… I’m sorry, the CEO of Zeenk, which is a direct-to-consumer analytics platform. Brian, how are you today?
0:00:25.7 Brian Eberman: Hi, good, thanks for having me on your show.
0:00:27.8 GF: Yeah, absolutely. So yeah, I always like to start with a little bit of background on you, your personal background and then how you became the CEO of Zeenk, and what Zeenk is.
0:00:38.7 BE: Sure. I can start with my own personal story if you want. So I’ve been a tech executive here in the Boston area for quite a while, mostly focused on data and data analytics companies, ’cause I have a PhD in AI, and started off actually in the speech recognition industry. I was in that industry for about 10 years before moving into the marketing space in 2006 with a lead gen company, which I ran for another six years. And then since then, I’ve run multiple consumer-focused businesses, and in every one of those companies, the acquisition funnel, how you deal with the acquisition funnel, where you focused for consumer marketing is really a key element. And my last gig before coming to Zeenk was with Thrasio, the largest aggregator of Amazon brands in the world, where we really saw the opportunity of trying to help these companies as a platform play. Thrasio was buying a lot of Amazon and some Shopify companies, where the founders had basically reached the limit of what they wanted to do as a management capability and a growth. Part of the reason they decided that they couldn’t grow anymore was they could no longer… They couldn’t really understand the business, both from a growth perspective and from an operating perspective. So with Zeenk, we’re focused not just on advertising and marketing, but ultimately on really being the data analytics platform for helping e-commerce companies run their business.
0:01:54.9 GF: Awesome. And so from what I gather, the platform offers insights into some of the key KPIs that a brand will be looking for: Acquisition costs, lifetime value, average order value, churn rates, all these things. But maybe can you give a little more insight on why a brand would find value in Zeenk and what they’re using it for to extract insights?
0:02:15.4 BE: Yeah, so the really big difference, I think, for us versus the other companies in the space is we really come at it from a customer analytics perspective, which means we think that brands… I was in marketing and product management for many, many years, and there are the five Ps of product management, and one of the keys is, understand your customer, understand what your customer wants, understand what the best messages are to reach your customer and understand which customers are worth the most to you, both in revenue and in cost of acquisition. So we take that individual customer lens and all the analytics we do for our brands, it shows up the most specifically in our ability to give companies insights into the lifetime value of each individual customer instead of a monthly cohort, and also give them a view of the individual cost of acquisition of each customer. And you see some really interesting plots and data when you do that.
0:03:12.0 BE: Everybody manages to the monthly number that they’ve set as a CEO, founder or maybe the finance team has set, in terms of a CPA they wanna hit, return on ad spend they wanna hit, but that average is hiding huge distributions of value, and we think when you take that lens, not only does it help your advertising, but it can help you think about your product retention. I was listening to one of your shows, Gen, on marketing, in terms of email, and he talked a lot about how you wanna make your email very specific to the customer. That’s exactly our view. You wanna think about, what’s the value of this customer? What products are they buying? Why are they buying the product? Are they engaging with you after they buy their product? And we help get you the analytics to that so you can craft your retention channels or your acquisition channels more appropriately to your customers.
0:04:03.7 GF: Got it. Okay, so if we’re talking hypothetical or if you have a case study, so actually, we had Leo Carrillo from Hair Craft on as a guest earlier in this…
0:04:11.6 BE: I love Leo.
0:04:12.6 GF: Yeah, yeah, Leo’s great. And he’s continued to expand his product line, but if we take that as an example, and so let’s say Leo has 1000 customers and they’ve bought like… It’s men’s haircare product. And so I’m gonna just play devil’s advocate or play the naive customer and assume, “Okay, so Leo’s got 1000 people who have bought and the lifetime value would just be say, $20, $40, assuming maybe they bought one or two products. So that’s what you would help identify, is, how much have they purchased? But then, are you also incorporating predicted lifetime value? Because I saw that you highlight AI as part of your background, and there also is an AI element of what your product offers, so the lifetime value is not simply previous transaction value, but you’re assuming a lifetime value of future purchases.
0:05:07.6 BE: Yeah, I think obviously you know that Leo is using our platform, so I’m not gonna reveal a lot about Leo’s business plan. [chuckle]
0:05:13.9 GF: Oh yeah, yeah, yeah, of course.
0:05:16.0 BE: But we had a long relationship with Leo, he’s a great guy, he has a great set of products. They’re on… So there’s a couple of things that are really interesting to them, first of all, they’re on Amazon and Shopify, and one of the things we do… And this is not an individual customer, I’mma loop back to that, but he got into the business, he started his business on Amazon, and then he went to the Shopify site to get more insights onto his customers, exactly the same conversation we’re having here. Understand which customers are… What are his customers like? What products do they like? What are they really willing to pay for it? And then he can loop back and use that information to effect his Amazon strategy.
0:05:52.8 BE: The way we play into that is those two channels really work together. You’re advertising against your D2C channel, but it’s generating sales on your Amazon side. Measuring that and understanding what that’s really worth to you, can be a significant effect on your overall mix and your overall profitability, and we are gonna provide that insight to Leo at Hair Craft. The second thing that’s on the D2C side that’s really interesting is, just look at it… Before we get to the advertising side, we look at the lifetime value, the predicted long-term value. We build a model based on all the purchases of what each customer is likely to be worth to you over the next three, six and 12 months, and whether they’re likely to churn. And we use a data science engine to do that, an AI engine to do that.
0:06:41.9 BE: We can then… Since we have it at the individual level, we can look at properties of those individuals. What product did they buy? What product did they buy most recently? What product did they buy last? What advertising channels did you get them with? How much of the website have they been engaged with? Where do they live? What devices are they on? Etcetera. So an interesting insight that we gave him probably nine months ago was, his most valuable lifetime product was actually his least sold product.
0:07:13.8 GF: Interesting. Okay.
0:07:15.8 BE: So I don’t know if they’ve made any changes to their marketing strategy, ’cause the D2C side is a little smaller than the Amazon side, but that’s a marketing with large kind of insight that could affect your overall positioning of products. And of course, since this is a hair care and now skincare company, great products. It’s a high retention business, it’s about retention, it’s about repeat purchase ’cause it’s a consumable. And the one that had this great… Was basically a higher consumable. People had to buy it more often, and therefore it had a higher LTV value.
0:07:53.0 BE: On the pure advertising side, the thing that’s really interesting, not just for them, but for any company where there’s repeat purchase or even first time purchases is, when you look at the AOV plus, the Average Order Value plus the future value, and you plot that against, what did it cost me to get the customer? Everybody’s saying, “All right, well, in January, I spent this many dollars, I made this many dollars, so my average is this. My average CAC, my average AOV, and maybe I’ll plot my AOV over time for the same companies and for the same people and I get an LTV estimate.” So that tells you how well your company’s performing. It tells you a little bit about what your margins are. It tells you what your cash flow is gonna look like overall on the average. But the average and the spread are very, very different things.
0:08:42.0 BE: So you can very easily have… And it’s true for Hair Craft, it’s true for other companies. You can have people that are worth very little… Aren’t worth a lot to you, but cost you almost nothing to get. So, they’re actually super profitable. But then you can have other ones where you’ve touched them two, three, four, five times and you’re just bleeding money. And that actually tiers out. So you don’t just have the low tier, but you have the mid tier, the upper mid tier, the upper tier. And every one of those tiers… Even though, “All right this guy’s worth 10x to me, but it cost me 50 times as much to get him to show up, and come back. So I’m actually losing money.” So we see two opportunities, really. The first one is, and this is where we’re working on some case studies, is to stop spending money on the places where you’re losing money. Cut as much of that as possible. And the second thing we see is, and this is true for a couple of Thrasio brands. There are people willing to spend a lot of money, they may have very different internal motivations for why they’re buying your product than your typical average customer, but they’re also gonna cost you a lot more money to get. So, but the opportunity there is… Could be very large. And so by looking at it, this is again back to the email segmentation, the… And People think about personalization as communication, it’s also about value measurement.
0:10:09.6 GF: Yeah. And can you explain that value measurement, that strictly quantified dollars of like say, lifetime value.
0:10:09.6 GF: Yeah. And can you explain that value measurement, that strictly quantified dollars of like say, lifetime value.
0:10:17.0 BE: Well, not just value of cost, you have to consider both value and cost. You could cost… You could be worth a ton to me. You buy Hair Craft product every month, you’re great. But every time I have to get you to come… I have to tell you again to come do it on Facebook. It costs me more on Facebook than the product’s worth to me, you’re a terrible customer. [chuckle]
0:10:40.1 GF: Right. Yeah. So now it’s starting to make sense in terms of the insights that the data would give. And data’s useless unless you’re actually able to make decisions based on this with the confidence that it will increase revenue and profits. Basically, the challenge that a lot of brands… That brands are facing now is uncertain which channel to attribute the revenue to, uncertain what the actual acquisition cost is, uncertain how it plays out between different platforms, say Shopify or Amazon and then spending appropriately for the value, the lifetime value of a company, so… Or, of a…
0:11:23.0 BE: Right, customer.
0:11:23.8 GF: Yeah. So you’re basically saying… Let’s say you have a marketing budget of $10,000, you basically just want to be most efficient by finding the highest LTV/CAC ratio, right?
0:11:36.5 BE: Correct.
0:11:36.7 GF: That’s where you’re zoning in on?
0:11:37.5 BE: Correct. Because that’s where… And you really want LTV to include your… The total cost, not just the revenue. But these are all D2C brands that we’re helping. They’re shipping something that costs them money, it has COGS, it has return rates, it has shipping costs. Those have to be accounted for in the value calculation side, on the value side. And then you want the highest ratio of that resulting contribution profit to the cost of getting them to be acquired and stay with you. So, we are compliment… So attribution, we do do attribution but really we don’t focus on that sign ’cause attribution is about, “Okay, I got this first piece of revenue, which of my channels accounted for it?” And it’s kind of a black art and it’s a little bit hard to do without really understand… Knowing anything about the views, which you cannot do today. No one can do the views except for Facebook and TikTok and those guys. They can’t do the clicks anymore. That’s why we have [0:12:42.2] ____ this disconnect and we can get into a long discussion about why this is. We focus on what you can measure, which is the first-party click data, what you’ve actually spent, and how you’re going to take all of your spending and attribute it to the ones who actually bought.
0:13:01.7 BE: And so we have a technique, a data science technique for doing that, that preserves the… Well, I would say the heterogeneity, but the equivalance of the various people. We assign the people that are most liked to each other. So if you think about it, let me put it this way. In a month, if I wanted to… If I was finance and I wanted to compute CAC, what I would do is I would take the total spend on all the acquisition channels and I would divide with the new customers I acquired. That’s my CAC.
0:13:29.9 GF: Sure.
0:13:30.8 BE: Super simple. Don’t need a platform, can download the data. I can do finance for that, right? So what we do is we say, “Alright, that’s the correct number on average. There’s no question that that’s the correct number.” But what we wanna do is say, “Alright, got a bunch of people who came that we spent money to get to show up, but they didn’t buy anything.” So, all those costs have to be assigned to the people who actually bought something, and we wanna do that in a way that maintains the… That basically makes all… Only assign it to people that are most similar to you. So, “Did you come on this amount? Did you come on the same day? Are you from the same device? Are you… Have the same behavior on the website?” There’s a bunch of parameters that we have that basically say, these two are similar. This guy bought, these guys didn’t bought. I’m taking the cost of those who didn’t bought and assigning it. And so that’s how we compute CAC.
0:14:26.5 GF: Got it. And of course, this is all overlaid with the challenge of the current tech landscape, iOS 14, and the Pixels, and kind of like disconnected data, which is exactly I think where it comes to the need for almost a plug-and-play solution, whether it’s you or some of the other alternatives on the market. Can you explain a little bit of the overview as you see it on the challenges that brands are facing now to actually get a source of truth or something that they can believe in? Because from what I’ve seen and read from other brands, every platform is trying to claim that they deliver this result. And so there’s like… It’s a little bit muddied, and so like, who to believe? Like you said, you see the top-line revenue. That’s what matters most, but kind of like a layer down, everybody is clamoring for…
0:15:19.6 BE: And the topline cost.
0:15:22.6 GF: Yeah, and topline cost, yeah.
0:15:23.3 BE: So let’s make sure we understand what actually happened and what the future trend is. So, I mean, the goal… Apple’s goal, and Google’s goal… And Google’s a little bit more late to the party ’cause they have all their advertising dollars on it, but if both companies feel like… If I’m a brand, I have no right to know something about that user deeply until they become a customer. That’s effectively their position.
0:15:52.0 GF: Right. Their activity outside of your site, you can’t…
0:17:14.0 BE: So Facebook’s pixel falls in that, TikTok’s, Snap’s, even GA’s falls in that category. And they basically said, “We’re gonna block that and we’re… And then the last thing they did is they said, “Anything… Any cookie that you drop on the client side, I’m gonna delete after seven days,” end of story. So that’s why Facebook’s attribution window is now only seven days. That’s why they made it seven days. So those are the things that are changed, and there’s like… No one can solve those problems. It’s just… That’s just the way it is. So what people are doing and what we’re doing is you have to do data science, and you have to do it at the click level, right? And you have to look at the clicks coming in, and you have to sort of build a model of, “Okay, I know that one came from Facebook ’cause it says so in the UTM parameters.” I know this one came… This one’s unattributed ’cause there’s no tracking tokens here. I gotta figure out whether it likely came from Facebook or whether it was caused by Facebook using some data science modeling. And we do that. I think that’s what everybody else is doing, ’cause in the end, that’s all you really can do. And you build correlative models to try and figure out where that traffic is actually coming from by looking at how much traffic they say they’re creating and how much traffic you’re actually seeing, right?
0:18:29.2 BE: And ultimately, that’s… These kinds of data science driven models that are in first-party data, which are going to be not specific to a user, but specific to a channel. In the end, what we’re saying is, this ad drove this much traffic. I don’t know who it is necessarily, until they buy, but I can say that I know that all the… That this ad, I saw this many people come in on that cookie, on that ad, and overall, we think it drove this much traffic in total, so it had this many clicks of your total clicks, right? That’s all first-party modeling, and that’s gonna be safe, and that’s gonna continue on into the market, and that’s where we’re gonna go. And the same thing is gonna happen in mobile. You’re gonna have to say, “Alright, on mobile, download-based businesses, how much traffic did I get to my page where I send you off to the mobile app? How many app starts that I get that were new? What’s the relationship between these from a data science perspective? I think that’s what’s going on. You’re not gonna be able to do direct tracking because the platforms aren’t gonna let you.
0:19:39.5 GF: Yeah, got it. So this is where you’re saying, if you can get down to an individual customer level, not only to know their purchase history, their value, but to know their source and the journey that they took to the site and to purchase. Those two data points are invaluable to making these decisions on how to market more correctly, how to spend more correctly.
0:20:06.5 BE: Right. And I’m saying more specifically, so we drop a cookie, it’s a first-party cookie coming from the website, from the company itself. So it has better properties than coming from a tracking pixel. We then look at the data and say, “Alright, we know you got 100 clicks from Facebook, ’cause Facebook says you did, on this ad.” We can see this many. [chuckle] So, there’s a bunch missing, we can figure out what’s missing and where it is in the missing clicks. ’cause I know you had 1000 people come to your website. And so we can say, “Okay, how many came from that ad? How many came from that ad? How many came from Google Search? How many came to Google Search because they came from Facebook or TikTok? That’s the kind of data modelling we end up doing.
0:20:52.3 GF: Got it. Now, if we talk… That’s on the acquisition side. Now, more of the extended lifetime value and focusing on retention and churn, which is something that your customers also focus on, to draw and take from. Kinda like at a high level of strategy, what are some of the effective ways to better brand… And again, a consumable like a haircare product makes sense besides maybe, “Okay, we assume that this hair gel is generally used within three months. So, three months from purchase day, we’re gonna send a reminder in the email, “Hey, do you need to replenish your order?” But it [0:21:24.8] ____ cares to know where you’re seeing brands that are succeeding, what strategies they are employing in order to improve their lifetime value.
0:21:35.4 BE: Right. So our analytics looks at the lifetime value, but not only that, but makes a prediction of when the person is most likely to buy again. And what is the likelihood today that they’re still a customer of yours, which means that they’re likely to buy from you again. So we compute a couple of things: Are you still a customer? What’s the probability you’re still a customer, and what’s your future value?
0:21:55.5 GF: How are you defining if you’re still a customer…
0:21:57.2 BE: It’s the inverse of the churn. So, churn is the probability you’re never coming back. And that’s what churn is by definition. So, one minus that is that you’re still a customer, you’re likely to buy again in the future. So the ones that are… Does that make sense?
0:22:13.0 GF: It does, but then if say I’m buying at a T-shirt… A store that sells T-shirts, I guess if I don’t buy… If you aggregate buys a second product within six months, if I don’t buy within six months, I’ve churned out, but if I’m within that six-month window, I’m still a customer?
0:22:32.5 BE: It’s actually more specific than that. We’re looking at each individual. Let’s take the haircare product. We know how often you buy, you as an individual. But not only do we know how often you buy, we know how often each individual buys, buy product. And so we build a model of, “If you buy that product, if you’re you and that product is the set, how do I get the max… The best fit against the entire set of consumers that predicts the likelihood of buy for every single individual consumer?” So, you go through your hair product fast, you’re likely to buy every month. If it’s six weeks out, you may just have churned.
0:23:17.3 BE: Another guy buys every three months ’cause it’s more sporadic. So, at two months, he hasn’t churned. So you need to know that. And so that data… And then you need to know the value, so the guy who buys every month is obviously worth more depending on the product, than the guy who buys every three months. So we give you that segmentation ability to look at that on a customer by customer basis. And that becomes an overlay to your best-of-breed email marketing strategy or any kind of communication strategy you’ve got. You should still do best of breed, personalized email kinds of communication, talk about the product that they like to buy. But now you can do an overlay that says, “Well, these guys are gonna buy anyway. Do I send them as many emails or do I send it less often?” Probably don’t wanna put in an offer. Probably wanna really think about an offer tied to recommendation instead of an offer tied to purchase. ‘Cause these are your multi-buyers really brand… They’re really loyal to your brand.
0:24:18.9 BE: Then you’ve got your mid-tier, which are the… Well, they’re probably gonna buy again, but you might need to incent them to do it again. That’s where you should really hit with the discount. And then you’ve got the low-end guys, like, they’re probably never gonna buy from you again. You should probably figure out a different product to send them from a couple times, and then just say, “Okay, I’m gonna stop sending them ’cause I’m gonna get counted for spam.” So that’s a separate segmentation than what product they’re buying, what are the demographic characteristics, what kind of messages do they respond to, that we see the two things overlaying together to make a best-of-breed email and communication strategy.
0:25:00.0 GF: Yeah, what popped into my mind is like an RFM segmentation: Recency, frequency, monetization. Which maybe in some ways is like a different approach to the same outcome, is like, prioritize those who are the highest value, don’t discount in order to get them back ’cause they’re gonna purchase again. And then you go down the line, I think RFM might have nine different tiers or something.
0:25:20.3 BE: This gives you more specific… Makes it easier to do the RFM, ’cause it’s just like, “I can just slice them pretty easily.”
0:25:27.0 GF: Yeah, I love it. I think the main takeaway in this, whether it is a direct-to-consumer brand or a SaaS brand, it’s… The foundation still applies. Understand the value of your customers, understand how much it cost… Took you to acquire them, create different segments. And then the strategy in which you are spending more, assuming you’re using paid channels would differ depending on the different segments. And then just being thoughtful in terms of how you’re going to retain them in an effort to reduce churn.
0:25:57.2 BE: Exactly, exactly. And then you can measure the results on our platform and see how well you did.
0:26:03.4 GF: For sure, for sure. Yeah, so where can we learn more about you and Zeenk?
0:26:07.5 BE: Come to Zeenk.com. We talk about the platform there, and we’re happy to do a 30-day trial and demo for anyone who wants to see more about the product. And we just love talking to brands and understanding how their particular issues are playing out. Every brand has a slightly different story and slightly different set of issues.
0:26:24.0 GF: Totally. Yeah, Z-E-E-N-K.com. Brian, thank you so much, I appreciate it.
0:26:30.2 BE: Great, great to meet you. Thanks for the discussion. 0:26:33.8 GF: And that’s the episode for today. Thanks so much for listening all the way to the end, we love you for it. If you found anything valuable at all or wanna share your feedback, please leave us a review on iTunes or wherever you get your podcasts. And you can also just drop us a line, firstname.lastname@example.org. We’d love to hear your feedback or suggestions, so I can cover it in a future episode. Alright, see you next time.