Add To Cart: Australia’s eCommerce Show

The 25% Rule: David Chinn on How Lexer Helps Retailers Find Their Best Customers | #567

• Nathan Bush • Episode 567

Share episode

David Chinn, CEO and Co-Founder of Lexer, knows better than most that retailers are drowning in data but starving for insight. Since 2015, he’s been helping brands like Cotton On and Rip Curl unify their customer data and use it to drive measurable growth. From tackling TikTok Shop duplicates to using AI for smarter segmentation, David and his team are showing what’s possible when retailers actually take control of their customer data.

Today, we’re discussing:

  • How first-party data is reshaping ecommerce in a cookieless world
  • The messy side of marketplaces like TikTok Shop and how to fix duplicate data
  • Why the smartest retailers focus on their top 25% of customers
  • Zero-party data done right (and when to ask for it)
  • How Lexer is using AI to turn insights into real-time action
  • What’s next for data-driven retail and where brands should focus in 2025

Connect with David
Explore Lexer

SMS us to request a guest!

Support the show

Want to level up your ecommerce game? Come hang out in the Add To Cart Community. We’re talking deep dives, smart events, and real-world inspo for operators who are in it for the long haul.

Connect with Nathan Bush
Contact Add To Cart
Join the Community

SPEAKER_01:

But then there's like there's nurturing. So in that bottom 75%, who are customers that look like the top 25%? How do I nurture them? There's probably the best time to collect information on a customer is in your order confirmation page. I never am I more excited about a brand when I've just purchased something and I see order confirmation. Let's load in transactions, every transaction you've got on all channels. Load in your marketing engagement information and put the tag on the website to collect that on-site browsing behavior. Now that's gonna get you to the vast majority of your use cases within the first one to two years. Start there.

SPEAKER_00:

What if I told you that one of the biggest untapped growth engines for your store right now is not a spin-to-wind pop-up. It's actually the data that you already own. Well, duh, right? Most people already know that. But the problem is, most brands aren't even using half of it to its potential. In this episode, we are going to dig into how to turn your customer data into a growth engine. And it's not as complicated as you might think. Today I've got with me David Chin, the co-founder and the CEO of Lexa. David and his team have built one of Australia's, and soon the world's leading retail-focused customer data and experience platforms. A CDXP, if you will. And it's powering today some of our biggest brands. So a little bit of background first. Lexa started in Melbourne and has grown into a global platform built specifically for retail. And their promise is to help brands understand, engage, and grow their customers through unified data, insights, and activations. That's a lot of buzzwords, right? We're going to break that down. They've worked with big names in Australia such as Cottonon Group and Hairhouse to really unify their data, power personalization, and drive customer lifetime value. On top of that, they've raised some serious capital. They raised$25.5 million in Series B, led by Blackbird, which have a history of backing some pretty good players, and King River, and has fueled product expansion, teams in multiple regions, and deeper integrations. Today, with David, my goal was to dive into helping you understand how to make the most of customer data. So we talk about what is the difference between a CDP and a CRM, which data capabilities are the lowest hanging fruit to move the needle for you, the difference between first party and zero-party data, and real case studies where a CDP would actually fit in the modern tech stack. And David shares some real case studies and insights from Aussie brands that are using Lexa. So if you have heard the word CDP thrown around and you're just not sure where that fits in the overall architecture for you, or when you might need a CDP, this is the episode to clear that up. A big thank you to our partners at Shopify and Clavio for supporting Ad Descart. You make it possible to bring all these episodes to life, and we are extremely grateful for your support. Enough talk. Let's get into CDP World with David Chin, CEO and co-founder of Lexa. David, welcome to Ad Descartes. Nathan, thanks for having me. Awesome to have you here. I've been a big fan of Lexa for a long time. I will admit I haven't been hands-on with the platform too much recently. But I think most of our listeners will recognize you, if not from Lexa itself, but from your merch. You know what I mean? You're famous for your merch, aren't you? I mean, it's probably not the thing you want to be famous for. No, we're really proud of it. We love what we've done with the merch. Yeah, it's like the rockstar merch of e-comm. But for those who have seen Lexa around, and no doubt most of us will, because you're at every every show and you've got a really great team representing you. I'm keen to understand from you. How do you succinctly state the problem that Lexa solves for retailers?

SPEAKER_01:

Yeah, look, I think actually, Nathan, our merch is a wonderful lead-in to that because the merch has a purpose. Part of our mission is to help our customers to become data rock stars. So here you get the metadata metallica, you know, Dart Punk, Data Punk. And so look, Lexa's mission at its core is to help all customer-facing teams within a retailer to understand, grow, and retain their customers without extensive support from IT and analytics. And so, you know, you think today in a in a modern omnichannel retailer or e-commerce business, you know, there's a lot of things that hold teams back from achieving that outcome. And so it could be, you know, data accessibility, it could be that their data's trapped in all the different point systems they use, or, you know, they may have a data warehouse, but you know, the IT team owns the data warehouse, and as a business user, you can't get access to the data warehouse. Yeah, it could be data quality. So it could be that you know the data's raw or messy, or there's some issues with the underlying data that make it hard to get the data, the insights you want. It could be that we were on this platform, we migrated to this platform, we only migrated some of the data over. When we migrated, there was some issues with it. Like there's those problems are really common. And when you're talking customer-faced, do you say customer facing or customer supporting? Customer facing. So I just think about all the teams who have a role to interface to the customer: marketing, e-commerce, retail, service.

SPEAKER_00:

Okay. I'm really keen to dive into this because I know customer data is just an area that we all know we need to get right, but it's not easy. It's so hard, especially the the goalposts keep changing. The tech platforms get more and more complicated. And probably the crossover between platforms, I think, is something that's really, you know, perplexing a lot of people at the moment around where does my customer data live? And I will say that this isn't a sponsored podcast at all. This is actually us having a conversation because you're an Australian-founded platform and you've been doing this for a long time. So that's the reason we're having this conversation because you guys are the experts in customer data. But to kick us off there, can you explain the difference between a CDP and a CRM? What's the difference? Yeah, we get we get asked this a lot.

SPEAKER_01:

Mate, I'm no history buff, but I always go back to like where do these technologies originate? And so, you know, CRM is a late 80s tech. It's designed to centralize customer data and interactions. It was built for B2B industry use cases first, primarily used by salespeople. I think one of the important things about a CRM is the data in the CRM was keyed in with the fingertips of the salespeople, majority. CDPs are uh like emerged in like early 2010. Definitely an evolution of CRM. So it serves the same customer-centric mission that the CRM stood for, but really built for B2C businesses with a stronger focus on marketing. So the data isn't keyed in by fingertips. Actually, most of the data is keyed in by the customer themselves, sitting in different systems. And the problem that it's trying to solve for is how do I bring that data together and make it functional for the for the business? And you know, one of the strong stakeholders of that is marketing.

SPEAKER_00:

So are you trying to bring together all the disparate systems that might be collecting customer data or interactions and bring them into one central place?

SPEAKER_01:

Your question's a really good one because I mean that that's the origin of where CRMs and CDPs come from. But like fast forward to today, you know, a lot happens in a decade in technology. And you know, I think for people looking at the CDP landscape and asking a question of like what's the difference between a CDP and a CRM, you probably have to first understand that there's a lot of different CDPs out there. And I try to make it simple for people, which is to say, look, broadly speaking, there's three types of CDPs. Yep. And they all differ based on the industry and customer they serve. So there's a infrastructure CDP, which is sold to tip technical buyers of enterprise organizations, and they're all about like data infrastructure. How do I, the pipelines that connect data into a data warehouse? And they will have the ability to do better.

SPEAKER_00:

Okay. So they're communication CDPs.

SPEAKER_01:

And they will serve typically the team that is doing communications. So and they think of data in its role to do better messaging. And then the third type, which is probably the most broad as it stands from like this is the most variants, are these industry-tuned solutions, which is a relaxer fits. So we picked an industry vertical, which is retail, and we said, well, how do we help solve the data problems in retail? And one of the things that we found was after we've sold the data unification and creating that single view of customer, we actually needed to go a lot further and provide lots of tools to those different teams for things like segmentation, insights, media activation, one-to-one engagement, in-store clienteleing and measurement. So when you ask that question of like what's the difference between a CDP and a CRM, I'm like, well, what kind of CDP? Because it they've evolved quite a lot.

SPEAKER_00:

They have, they have. And even, you know, we we alluded to it as like platforms have evolved around us. So if I take the most common tech stack, I think, in e-commerce at the moment, you have a Shopify, you have Clavio on top of it. Clavio is sending, obviously, Shopify is taking orders. Clavio is then sending communications, marketing communications via email, SMS. They've also now got CRM capabilities in there as well from a B2C capability perspective. At what point or what problems evolve from that basic tech stack into going, actually, I need a CDP over the top of all of this?

SPEAKER_01:

Yeah. Look, I'd say that the we don't replace any of that technology. We sit alongside it. So I just architecturally we will take data from those systems. We will then do a lot of enrichment on the underlying data in those systems. Because if you just think about email engagement, there's like clicks and opens and buy campaign. There's a lot of lot of really granular event data there to make use of. So we take all that data, enrich it, make it easier to access, and then connect it actually back into a lot of those systems. Okay, so it's two-way. Yeah, it is two-way. I think your question's a great one, which is like when would that tech stack, a customer using that tech stack, when would they first think about reaching out to Alexa or adding Lexa to their architecture? There's probably two primary like lightning points that we often find ourselves speaking to customers about. The first is when there's a need to dramatically improve their understanding of the customer. Both Shopify and Clavio as tools, amazing technology, have done an incredible job to scale, but not a lot of them are great with data discovery to actually tell me something about my customers. They're great at like operational reporting or understand the performance of your campaigns. But if you want to start who are my best customers and why, what are the purchase journeys that deliver the best long-term customers, you know, looking at my customer base, who's at risk of churning and why, those types of questions that they just haven't been built around that type of upfront discovery. The second one is there's usually a desire to, you know, be more personalized or move beyond a couple of journeys, like a welcome journey and a card abandonment and a win-back journey, and then a certain dissatisfaction with blast messaging, every customer with the same new product launch and promotion. So there's a desire to be more personalized beyond a couple of basic journeys and to move away from blast messaging. And then they come to us and we get involved with the strategy of how you do that. And then our technology helps underpin the ongoing execution of that.

SPEAKER_00:

Okay. Okay. Do you also find that people need to go to a CDP when channel strategy gets a bit messy? So, for example, opening stores, so you need to pull offline data in, or potentially expanding into different territories where you've got different stores open for different regions and you need to combine that data?

SPEAKER_01:

Look, the second, like channel complexity is a fantastic driver for a CDP. So, hey, I've got using this point of sale system, you know, doesn't do great with data exporting. I've got e-commerce transactions here, and you know, I need to unify these different data sets together. They might have different identifiers in them to match customers between them. All of those use cases are a fantastic CDP use case.

SPEAKER_00:

And you mentioned it before, the single point of truth. I think that is something that all retailers are striving for, that customer single point of truth. Is it a realistic goal to get, and I'm not even lobbing up a question for you to make to kind of justify your capabilities, but I'm seriously wanting to know whether that's an absolutely realistic goal to have a single point of truth that captures obviously purchase data, behavioral data, psychographical, demographical data all in one place, or is that kind of a phased approach that you find most people do over time?

SPEAKER_01:

Oh, hey, very, very real. Like we do that for many, many, many brands. I think look, you took like omnichannel, obviously, online data is usually pretty good. I think the big determinant of success with the retail or offline data is just the append rate of customer data to the transaction. And you know, we're seeing What do you mean by that? Well, okay, so I go into the store and pick up a pair of jeans and buy them. At the checkout, they say to me, Are you an existing customer, yes or no? And I like, is there a strong incentive for me to say yes? Some of them are using e-receipts. Hey, can I give you give me your email address so I can email you a receipt? It's not a super compelling call to action for a$50 garment. Loyalty programs are obviously, you know, doing a pretty good job of giving the consumer a reason to identify themselves at the point of purchase. So just as long as the append rate of customer information to that transaction at the point of sale is decent, then it's fine. But Nathan, you know, the thing that we're seeing now, which is a way bigger problem than the append rate of retail transactions, it's the emergence marketplace. So Amazon and TikTok shop in particular offer the capability for them to fulfill the order on the retailer's behalf. But then they don't want to provide the full customer information. And so especially when so there's a type, you know, it could be Amazon 3P or TikTok shop where you fulfill your own orders, they will provide you the customer's name and delivery address, but they will not give you their email or phone number. So, and for most of like e-com brands on Shopify, those transactions are coming back into Shopify for the fulfillment pathway. And so create a duplicate customer record. And so we're seeing like we had a customer in the US, a beauty brand, one of the celebrity-backed beauty brands, who, from a standing start in their first year on TikTok shop, did 43% of their new customer acquisition on TikTok shop. But a significant proportion of those customers were actually existing customers in the e-comm file. But because they had no email or phone number, they were coming in as duplicate records. So we do smart stuff like, you know, using postal address validation software, we resolve the address to a barcode and we combine the customer's first initial and surname to the delivery point barcode across the e-comm file and the marketplace transactions, and we then are able to unify and avoid that duplicate. So retail's easy, marketplace is complex, but you know, that's what we do kind of every day. And so we've developed some pretty smart processes to make sure that we can bring the data together and avoid duplicates that hinder your insight. Because that, like everyone's focused on their order frequency and the lifetime value. And when you start introducing these customers that already exist, it increases your one-time buyer rate, it lowers your customer lifetime value, it completely pollutes your understanding of recency frequency and monetary value. Your attention looks terrible now because those customers that are shopping in TikTok and might not be shopping on your site now suddenly start to age out, look like they're lapsed.

unknown:

Yeah.

SPEAKER_00:

And I think too, like everything that I'm seeing with TikTok shop and speaking to people about is that it's a brilliant acquisition platform, but on that first sale, you rarely make a profit. It's not cheap. So it's about getting that return. So I'm assuming that by having a high match rate to existing data and being able to remarket to that customer to get that second, third, fourth purchase, it actually becomes viable long term.

SPEAKER_01:

Yeah. So we could say, hey, that customer that just placed an order on TikTok, they actually match to someone in your e-com file. Let's kick off your next order journey, which is maybe a replenishment of the product they bought, or a cross-sell for a new message in owned channels where you've got the opt-in because they're an existing customer, and we'll just move them into that channel. And then if we're going to do any audience-based activation in that channel, suppress them out of that channel, knowing that we can reach them in owned.

SPEAKER_00:

Yes. Okay. Are there any other fields or data that you use to match customers up to remove those duplicates? Do you get into the technical areas like IP addresses, devices, all that sort of stuff?

SPEAKER_01:

No, look, before Lexa, I I ticked over to a decade at Lexa last week, but before that I did a decade at Experian. Okay. And we did a lot of kind of data quality and data matching solutions for marketers. And look, I I just think that the downstream impact of a false match is so much worse than a small percentage of duplicate within the data set. Anything that is not deterministic, like e even leveraging a third-party identity graph. I mean, we've looked at that extensively, and there's obviously some error in that data. And so just always lent on a very, very comprehensive match, deterministic match across all of the provided identity fields. So name and address, you know, name and phone number, email address, any customer IDs coming from those source systems, we'd be pretty focused on those. Which gives you, you know, when you blend them all together and have a matching logic across them, it's it's a pretty comprehensive matching solution.

SPEAKER_00:

And I don't expect you to give away your matching logic because it's part of your IP. But for retailers out there knowing, well, how do I get certainty? Is email to email enough, or do you go on multiple fields to get that certainty?

SPEAKER_01:

Yeah, we we we go on multiple fields because you know, different systems provide all provide all the required fields. And it was fascinating. You just have to move to America, get a new phone number, and to realize how quickly they recycle phone numbers in America. I I received for months phone calls and phone calls from the previous owner of the mobile that I'd had. You know, people look at a mobile phone number and think, hey, it's quite a deterministic match key. It's obviously you can't get a false positive, but numbers recycle. And so we combine features of the customer's name and the phone number to avoid the error rate that comes from the numbers being recycled. Whereas email's different, because and phone numbers are good because people typically have one phone number. But email's good because you know, people they're very unique and don't get recycled, but people have lots of them. Yes. And so, you know, we see customers frauding the new customer sign-up discount as an example. They give a new email address, and email's the primary key, but when they make their purchase, then they create a phone number. And so unifying across phone number will then have profiles with multiple emails.

SPEAKER_00:

And do you have customers using that matching to kind of stop people claiming that new customer discount?

SPEAKER_01:

I would say it's probably less about stopping them from claiming the discount, but identifying where, if that's a problem for their business, like the frequency that it happens.

SPEAKER_00:

Identifying the tight asses in their customer set.

SPEAKER_01:

Yeah. And then they might, do you know what? They put them into a because we look at discount rates, they'll put them into a discount group and probably be a little bit forthright with them when they're on promotion.

SPEAKER_00:

Yeah. Okay. Do you see any new unique identifiers coming in in the next few years? Because we've relied on email and phone number for so long, right? Like it's just been something we've done forever. Yeah. And when you really think about it, nothing's changed a hell of a lot. We kind of went down the path of, like you said, with Experian, all the behind-the-scenes tech identifiers, but with new privacy concerns that's kind of become a little bit gray area. Do you see any other identifiers ways that we're going to combine this data?

SPEAKER_01:

Oh, look, I mean, there are third parties that this is their core business. And so you provide them the customer information, they will then give you a synthetic ID that resolves back to their master list. But again, there's like a bit of error there. And the hey, do you know what? If you're this does not rank, like if you're robustly using name and address, phone number and email address, and you're a retailer that turns over less than$500 million, you've got way bigger issues than this.

SPEAKER_00:

Yeah, that's very true. I love where you went with the insights that your customers are getting from your reporting and insights features. Can you give us some examples of some retailers on the platform and maybe some of the unexpected discoveries that they've made from using this combined data source?

SPEAKER_01:

Yeah. Yeah. I could I we could do a whole hour just on this.

SPEAKER_00:

Go on, give away all their secrets.

SPEAKER_01:

Yeah, yeah. I won't name names, but I'll uh I'll tell you where where this stuck. I think the first thing is this we just see a statistic play out at such a high frequency across our clients that it's I'm always just I I see these stats come out and I'm like, wow, another one, which is the top quarter of their customers will contribute two-thirds of their total historical sales. So 25% make up 67% of sales. And you know, wherever you see that insight within your business, now where it changes, if you're higher price point in luxury, you'll get to that Pareto 2080. If you are a low-priced product where customers only buy one of, think like phone cases, something like that, you know, you're obviously not going to have that. You're gonna be like top 25, might make up 30 or 40% of your sales. But by far and away, like the vast majority of our clients sit in that 25, 67. If that's you, well, like that should dictate almost all of your customer strategy. So you should be really focused on who are those top 25% of customers and how do I acquire more of them? Because the bottom 75% that makes up a third, usually there'll be a six times variance in their customer lifetime value between the top 25% and the bottom 75%. And so many brands today are using Meta's dynamic ads for broad audiences, which is such an amazing and powerful tool optimized around first purchase ROAS, but it's not optimized around customer lifetime value and repeat purchase. So there's usually a like a some of your product catalogue that is very appealing in an ad to a buy, but it actually attracts a high volume of one-time buyers. Whereas focus on the items in your product catalogue that you know attract a greater proportion of repeat purchasers who spend more, you know, that often has a really transformative effect on revenue growth. But then there's like there's nurturing. So in that bottom 75%, who are customers that look like the top 25%? How do I nurture them? There's I just always think about like just economic models of investment. You know, if you've got a hundred units of marketing dollar to spend, how are you spending that across your customer base? Are you allocating the spend evenly across all the customers? If you're doing that and you have that 25, 67 split, it's crazy. You should be spending way more of your budget where you make your revenue. And so there's this just some simple things like that that early on can deliver really big impact.

SPEAKER_00:

And say you've got retailers who have found their 25%, and that's their their gold pile there. Are they uncovering new insights around behavior or preferences or maybe even intention signals using the data that you're pulling together?

SPEAKER_01:

Yeah, like lots of stuff. Like so we we have third-party data from Experian in the product, their household segmentation mosaic. And so often we'll see a SKU in the more affluent mosaic groups as an example. Or if they're like a younger, faster brand, you know, maybe it's the affluent inner city groups that we'd see. And so one of the things that we'll look at is in your top 25%, what's the distribution of those groups versus the population and how does it index? And then from there you start to get a like a view of, hey, my best customers look like this, think like this. And then, you know, as you get to your marketing and your creative, well, how do I best speak to that type of customer and you can tune your marketing towards them?

SPEAKER_00:

Okay.

SPEAKER_01:

We look at like product journey, mate the life cycle, like what product do they start with? What do they buy next? How frequently do they come back? Like all of that type of analysis when deployed at that group is really eye-opening.

SPEAKER_00:

Ever tried saving money on software and end up paying for it with a whole bunch of headaches? That was the case for WaterDrop, the fast-growing hydration brand helping Aussies drink more water. And these guys hate headaches. They are the enemy of good hydration. WaterDrop launched in 2023 and quickly hit over$1 million a month in sales. But when they tried to cut costs with a cheap SMS provider, results dried up quickly. So they brought SMS into Clavio. Everything changed. Deliverability bounced back to nearly 100%. Retention increased by 79% year over year. E-commerce revenue grew by 60% quarter over quarter, and they saw a significant boost in loyalty conversions. If you're ready to streamline your stack and stop leaving revenue on the table, visit Clavio.com forward slash AU to see how brands like Water Drop, scaling smart, and staying hydrated. And are you finding that your customers are creating segments within Lexa that they pass through to other marketing platforms such as Clavio or Meta? Is that how it works? Is that you keep your segments in Lexa and then use them as your source segments and then they go out to your other channels?

SPEAKER_01:

Yeah, that there's two ways. That is a very common pathway. So they build segments in Lexa. The segments are dynamic. So as data comes into the platform, all the segment populations rechange, and then we can activate them out to those paid and owned destinations and automate the updates. So as something changes, we can add them or remove them from the audiences. So you can make that the start of your journey or an ongoing campaign and remove them once they've taken action. Or the other way is we can actually just append information to the profile within that system. So Clavio, we can update the customer properties and append information there so you can kick off your trigger in Clavio leveraging one of the fields that Lexer updates on the customer's profile.

SPEAKER_00:

Oh, that's cool. Because then, yeah, you've got that full view. So then you can, I can imagine not every team wants to go into a CDP every time they discover data. You might have teams who just want to play in Clavio or teams who just want to play in Meta and let them be there.

SPEAKER_01:

Yeah, so we can sync information back on the profile or the raw events back into those systems to take action.

SPEAKER_00:

Okay. And you mentioned the third-party data that you're pulling in from Experian. Are you finding that your retailers are pulling in other third-party data at all, connecting to other sources to enrich the data that they've got? Do you know what?

SPEAKER_01:

There was a time there that that was a lot more common. People had their kind of desired source of third party. Um I would still say, like, just generally speaking, there's not that much third-party data available in Australia. There's a lot more in other markets, like in this in particular. So, but no, we don't see a lot of retailers bringing like extensive third-party data sets to the table and wanting to load them in Lexa. And we've kind of done the hard work for them to make it available in Lexa so they don't have to do that. Yes. Kind of one of the one of the value propositions.

SPEAKER_00:

Do you find that's because we've moved to a model around customer data where we're trying to keep it as lean and necessary as possible? Like obviously, there's going to be a lot more compliance that we've got to abide by. It's not here just yet, but it's heading that way. So you want clean data sets and using only the relevant information that you're actually going to use.

SPEAKER_01:

Yeah, it's definitely a byproduct of brands extracting more value from their first party data. I think third party was an easy proxy of like, I want to know something about my customers, but my customer data is too messy and hard to do it with. So I'll leverage external sources. And as it's become more easy and available to leverage your first-party data more, there's still I think that's kind of been at the expense of some third-party sources.

SPEAKER_00:

What first-party data do you think most retailers aren't making enough use out of?

SPEAKER_01:

Uh just transactions. There is just so much value in the transaction. You know, the if just think about the parameters of recency, frequency, and monetary value alone. So how often is the customer ordering? How many times have they ordered? What's an aggregate of their spend? Just those variables have so much inherent value, but there's quite a lot of IP and how to use them. And then the product that the customers purchased and looking at the sequence of the orders. So what product did they buy first? What did they come back and buy second? What was the time interval between their first and second purchase? There's just huge amounts of value there, but it's obviously it can become overwhelming really quickly. So finding the right level of aggregation to layer up to and picking what journeys you want to get into first, you know, there's a place to start there where that can be less overwhelming. But there's just huge amounts of value there.

SPEAKER_00:

Does Lexa stray into product insights as well? So obviously, yes, customers like this also bought this. But if you're going at it from the other way around, so say you're launching a new product or you want more insights on this particular product and the types of customers that bought this product and then may have bought another product or researching other things or time to purchase from first visit to last visit. Can you provide product-specific insights as well? Absolutely.

SPEAKER_01:

Yeah. I think the merchandising and product design teams love Lexa. And the reason being a lot of the reporting that they get is on sell-through. So how many units do I have? How fast are those units depleting? They don't get a lot of information on I've got a thousand units of product A. Was that bought by a thousand individual customers, or did a hundred customers buy 10 units each? Now, obviously, that's an extreme, but that's a pretty meaningful thing, is how you think about that product. It's actually a highly demanded product to a niche segment of customers, or is it a widely consumed product by a broad number of customers? That makes you think about like, what do I do about category adjacencies? How deep do I go within that, like, or how deep is my assortment within that category? There's a lot of insights and decisions that get made from that that you don't get from just looking at sell through. And so we're Really good at not only do we have this view of the customer in Lexa, which is like an enriched view of summarized all of that raw data into structured attributes, you can also analyze all the raw event level information.

SPEAKER_00:

Yeah, that's cool.

SPEAKER_01:

You can do things like say, hey, show me customers that clicked on that email campaign for brown handbags. How many of them actually went and purchased a brown handbag? And getting into the actual like that kind of that product level view to validate, well, that campaign was seen as a success, but it was designed to move product A. Did it move product A? Those types of things are really easy to do in Lexa.

SPEAKER_00:

Are you helping retailers with attribution as well? Because I think everyone was hoping by now we'd have the secret answer to attribution and that we'd be able to tell exactly what channels impacted what. Are you seeing retailers use it for tracking customers with how many touch points they've had before a purchase and attributing success to certain channels?

SPEAKER_01:

Yeah, this has been a hotbed for us in our product roadmap, whether we dip our toe in this field. And we've made a pretty deliberate decision to not go deep into the attribution space. I feel like today our value chains are already really broad, and that's going to be a product set that, you know, and there are some emerging vendors that are doing a really good job at gaining prevalence in the attribution space. Where we stop with attribution is I kind of feel like brands are drowning in campaign reporting in specific channels. But there's a real absence of measurement at the segment level and at the total customer level to understand, well, those 10 campaigns, if I add up attributed sales across all of them, it's double the total sales that I've got anyway. But if those can't if those metrics there are firing, is it actually influencing the core segment metrics? And so brands that are doing, I think, a really good job are the ones that like master the customer life cycle. They've got in Lexa, you know, they we've built an enhanced RFM model for them. They're obsessing over the movement between like dormant and active segments. And then they're they're running a campaign and they're seeing the migration of customers across the more favorable, less favorable segments, and they're saying, you know, is that the campaign that's doing the heavy lifting for us? And we can answer those questions, and then they're thinking of it that way.

SPEAKER_00:

You alluded to it before, but it feels like in the last six months, everyone's discovered buyering shops, how to grow brands around actually, no, for us to grow, we need new customers. We can't just service existing customers. And, you know, it's a lovely concept when you think just treat your customers right and we'll have a business for life. But everyone's like, actually, no, we need to, we really need to invest in finding new customers constantly. Do you think that retailers are spending enough time finding new customers at the expense of potentially reactivating ones that are already sitting in their database?

SPEAKER_01:

Yeah, we've we've definitely I think just the current economic environment has put a lot more attention on existing customers than new. So ROAS got increasing, budgets decreasing. So I if you had asked me that question, you know, three years ago, I would have said brands are nowhere near paying enough attention to their existing customers and you know they're way too focused on new. But I I think the balance has shifted and brands are doing a uh turning their attention through necessity on, you know, how do I maximize the revenue for my existing customer base?

SPEAKER_00:

Yeah. We had a really great webinar with Swim and Clavio and Convert Digital two weeks ago. And it was all around finding those intense signals from customers. So essentially zero party data. So whether that's from quizzes or whether that's from what you're looking at on site or all those intense signals before you even enter some identifying data in, are you finding that zero party data is actually being utilized by your retailers?

SPEAKER_01:

Yeah, we I mean, maybe it's like just definitionally, we think of the difference between zero and first, zero being information that customers volunteer directly to you. So that's me putting my hand up saying, here, have this information about me, whereas first party data is information the brand observes about the customer during the delivery of their service to them.

SPEAKER_00:

Okay.

SPEAKER_01:

And so maybe a way to think about it is if you were to ask the brand, does the customer explicitly know that you have this information? Like, did they give it to you explicitly or not? Usually that's like where you find the division between zero and first. So, yes, like first zero party, I like I've got two use cases because we we built a survey tool into Lexa because we were so excited about zero party data.

SPEAKER_00:

Yeah.

SPEAKER_01:

And so this whole thinking was how do you put the customer in the driver's seat of your relationship? Don't guess, just ask them. Let them tell you what's important to them and then deliver that to them. For example, if you're a brand that services multiple customer need states or hobbies or interests, we worked with Black Diamond. You know, they've got trail runners, climbers, and backcountry skiers. So in the email like data capture form, there's just a simple quiz there, which is, or question, which is what are you interested in? In the welcome journey that gets triggered, it determines whether your creative is focused on a guy with an avvy, like a backpack and an AVI beacon and 120 mil wide skis, or whether he's in a harness and carabina climbing a mountain. And so there's just like simple things like that. And then we would leverage that data through all of the, hey, we've got a new product within that category. Actually, are they interested in climbing or not? Are they interested in backcountry skiing or not? And then we would start to organize the content pillars around those.

SPEAKER_00:

I dreamt of that capability back when I was at BCF and you know, boating, camping, fishing was three very clear segments. You know, when you're saying that exam, I'm like, oh that seems easy now.

SPEAKER_01:

Yeah, one insight I would just share with your audience that I think is really important at the time of collecting zero party data, you want to do it at a time that the customer is there most engaged. So like the idea of sending out an email to customers to get them to give you some, like stop there what they're doing to give you some information. I don't think that's probably the best use case of resource because you know, you're gonna get X percent open, X percent click, Y percent form completion. Before you know it, you're at like fractions of a percentage of your file. But like probably the best time to collect information on a customer is in your order confirmation page. Sorry. Hey, I've just bought, I never am I more excited about a brand when I've just purchased something and I see order confirmation, I'm thinking, oh, I'm about to get that product, I'm really engaged. One of the use cases we've seen really effective, very effective there is understanding purchase occasion. So if it's a like what is the occasion that you're purchasing this product for, that can unlock recommendations into your catalogue of what other products, or it can unlock opportunities for you know future creative personalization.

SPEAKER_00:

What does that actually look like? So sending an order confirmation, am I putting a link in there to run a competition or am I asking for just information because they're nice?

SPEAKER_01:

Yeah, it could be I press or like submit or confirm the order, and then the page that pops that says, you know, the order is successful, you iframe the survey into that form that says say your order's committed. By the way, you know, what occasion were you purchasing for? Or you know, that's like because you get 100% of eyeballs are on that page, not having to market to them to drive them to that page, which has an incredible attrition, or paying for that. And so just making use of the eyeballs when they're there.

SPEAKER_00:

What about businesses who have a high gifting percentage? So you're obviously dealing with two customers here, the giver and the recipient. Are you capturing both as customers within Lexa at that point?

SPEAKER_01:

It's very rare that the person purchasing will then ship product to the like in our customer base. We don't have many florists. That'd be a really big one. But we would absolutely be able to create the profile for the recipient. The challenge is getting the marketing opt-in consent from the recipient. You haven't had a direct touch with them. But this idea of again, you just came up with a great was this product for you, if you've got a highly critical category, did you buy the product for yourself or for someone else? We have a gender inference model that's a neural network we've built that's trained on customer first name in Australia and the US and kind of those Western societies, where you have a very, very high accuracy rate to kind of upwards of 85, 90% of names, but the Sam's and Alex's, you know, the true Jesse's. Yeah, too hard. So we do a lot of like offer we have customers where it's a you know a predominantly menswear brand, or 100% menswear, and we look at, you know, the the gender of the buyer of the product, and they might have 30 or 40% females buying the products. You know, then we enroll them into a gift giving campaign around Father's Day and Christmas, and you know, we start to speak to them in that way.

SPEAKER_00:

That's smart. That's really smart. Just to loop back quickly to the zero party data, are you usually just bringing customers into Lexa once you've got that firm identifier? Or are you say I go on and I do a survey, but I don't give over an email address, for example, or do a product finder or something along those lines. Is that coming in so that when I do actually make an order or I sign up to a newsletter or something like that, I can backtrack and get that earlier data?

SPEAKER_01:

Yeah, we mostly see it less with zero party, more with like um website browsing behavior.

SPEAKER_00:

Yeah.

SPEAKER_01:

So we made a decision a while back that actually we we're looking at right now around, you know, that keeping a record using the Google ID, keeping it a record of those browsing sessions. And then when we see you have a source of like an identifying email, a phone, an address, one of those, that will backpopulate your profile with all your previous browsing. We don't do that today. But today we rely on having that kind of source of identifier, a sign up to the newsletter, a transaction, uh a clerk completing a survey. But it is definitely one of the items in our product roadmap.

SPEAKER_00:

Ever dreamed of turning a weekend passion into a thriving business? Girls with Gems did exactly that. What started as a mother-daughter duo hosting jewelry parties at home is now a bustling boutique and a booming online store. They've helped hundreds of women feel confident, stylish, and part of the sisterhood to become, as they like to say, one of the girls. And with Shopify, things are moving faster than ever. In 2024, their online sales revenue grew by jaw-dropping 107%. But it's not just about the sales. Things feel smoother behind the scenes too. Efficiencies have soared. Huge sales events are now prepped and scheduled ahead of time, and integrated automations mean that VIP customers are spotted instantly, triggering exclusive offers without anyone lifting a finger. Just smart tools, stylish growth, and one sparkling success story. To read more case studies like this, visit Shopify Case Studies online. I can imagine your product roadmap must keep you up at night when you don't know what regulations and what changes are coming, because both you've obviously got government regulation that's up in the air, but you've also got platforms like Google and Facebook and everyone else putting up walled gardens around data. You must feel like you've got a roadmap that you're kind of in control of, but anything could happen at any moment. Yeah. It's funny.

SPEAKER_01:

I feel like, well, one, we manage integrations into hundreds of platforms. And so I've got a whole team of people updating the schemas. So, hey, by the way, that's changed, and we react to it and update the schema. Oh, on the privacy world, you know, I feel like we've we've done GDPR, CCPA, the Australian privacy principles. We are pretty well dialed into any privacy change, frankly, from now. There's not a lot that keeps me up at night around changes to data. Yep. There's a bunch of other things that keep me up at night, but they're not.

SPEAKER_00:

What keeps you up at night at the moment then?

SPEAKER_01:

Oh look, we're just so obsessive around our AI roadmap.

SPEAKER_00:

Okay.

SPEAKER_01:

The pace of change there is just so intense. That picking the things that you build on and build strong IP in today that might become a feature of one of the LLMs tomorrow. Like, you know, I just think of this analogy. There was a company that was the flashlight app, and one day that became part of the iOS. It was just the flashlight was there and overnight their business was gone. You know, we can see OpenAI's just released their agent tool, and there's a bunch of products out there that are like N8N and those things that you use to build a genetic workflow that's now just a feature of a product. So we are just really thoughtful right now around how we help our clients maximize the value of those LLMs. What are the unique territories that we can use to accelerate the value that our clients get from them?

SPEAKER_00:

I can imagine, though, that there's also the opposite side of that in that by being an independent place where data goes, not owned by any one platform, you give that independent view and independent storage of data, which can play in your favor.

SPEAKER_01:

And do you know what? Very, very deliberately, we've made a decision to leverage the LLMs that enable you to retain your data within your, like we call it our red zone. So there are a lot of products out there that are wrappers on ChatGPT as an example that are sending data to ChatGPT to power their service. And, you know, we're seeing the horror stories of customers search for you can search for your name in chat histories, Google's indexing chat histories with your name in them, the that are your chat histories. We made a real deliberate decision to deploy LLMs like Anthropics Claude that we can run on our infrastructure so client's data doesn't need to leave our infrastructure to power those use cases.

SPEAKER_00:

Yeah. And when you think about it from a retailer's perspective, so whether you're a marketing manager, head of e-com, how do you think AI is going to change the game for how they use customer data?

SPEAKER_01:

Yeah, I think it's really exciting because you know, just the automation of tasks is there's like a first pass. So we introduced an AI agent into Lexa about 18 months ago called Lexi. Our first use cases were the automation of insight creation. So wherever in any of the interfaces in Lexa that we present customer data and information on a segment or the total population, Lexi lived there and Lexi has access to context about the company, the individual that is making the query, so their role and their seniority, and businesses' objectives, their key objective. And so the insights would be quite contextually relevant based on the data you're looking at in the user and the company. We used it for anomaly detection. So hey, you're presenting some, looking at some data, it's looking at the underlying data and telling you whether the data's sound to make good decisions or whether there's gaps or holes or potential issues in the data you're looking at. We also created a one-to-one messaging tool where you could create a segment and then craft a messaging brief to Lexi, and Lexi would write true one-to-one email copy to those customers. We've got customers that are generating like seven or eight dollars per email sent from that product. It's quite quite crazy. Like all of those products are about efficiency. So there's this paradigm of like, you know, how do I get an insight into identify a change in performance that I should be aware of, both positive or negative? So have the AI reading my reports and telling me what's changing. Then if I want to do something like workflow related, create a segment, build a report, activate an audience, like that should just be a natural language interface to tell it to do that. I shouldn't have to learn your UI and your data structure.

SPEAKER_00:

And do you think we'll actually even have data reports in a few years' time? Like, is there any reason to have reports when you've got a live engine that you can go back and forth with to find specific answers?

SPEAKER_01:

Yeah, I think there'll be a hybrid. And there's some operational rituals where teams look at reports together that I think are really valuable, like a cadence around, hey, here are my key metrics and movements, and as a team, we'll look at them together. There's some accountability that certain stakeholders get in an executive group to being held account that we're going to look at this, and structured reporting is a great thing there. But there's a bunch of stuff that's happening that the reports miss or that, you know, don't show as being a really significant change. And I think the field that's most exciting is the something's happening that's adverse. It's a strategic recommendation of what you need to do about it. And by the way, click a button to do it. And that's like activation to that's where I think it gets really, really interesting and where we're investing all of our kind of product resource right now.

SPEAKER_00:

I think that makes a lot of sense too, because most of the problem, you can have the best customer data in the world if you don't know the right questions to ask of it, it's useless to you. Do you actually think like there's that real barrier that for a long time we were sending kids off to be developers and data analysts? And do you actually need a lot of skill now from an analysis perspective to make the most out of customer data?

SPEAKER_01:

I it's I mean, as the I I'm a father of a 14-year-old and 13-year-old girls, and I think to myself, what will I encourage them to study at university now? And I the answer is I have no idea. That's completely up-ended education. It's one of those hard things to answer because you know, when you come from a deep area of subject matter expertise, the questions you ask are tuned in a certain way. So I still think there is like a that foundational understanding is pretty important, but they will significantly democratize access to information to a broader group of people. So I think there's going to be like, you know, the top 80 use cases are probably satisfied by this, you know, pretty easily.

SPEAKER_00:

Yep.

SPEAKER_01:

You know, you might have a team of specialists that'll take you a little bit further.

SPEAKER_00:

But it's not hidden in a black box by some gurus out the back that you need to ask for answers and have them delivered to you in a few days' time.

SPEAKER_01:

But I but Nathan, that's actually because a lot of this stuff is black box. And that's, you know, I I think understanding where it got its data from. So I think that the providers that are doing a really good job leveraging the automation capability of AI, but are giving the user information to this is where we got this insight from, so you can go there and validate it for yourself. So it's not black box.

SPEAKER_00:

Yes, yeah, yeah, I get it. So it's accessible but secure. I just want to wrap up here because there is a stat from Gartner, and I take them with a grain of salt, that 47% of CDP capabilities are not utilized by teams. If we've got people now listening to this and going, I see where a CDP could play in my business either now or into the future when we hit those next milestones. If you were to sit down with them and say, all right, for you to get the most value out of a CDP on top of your current tech stack, what would you encourage them to do in those first 30 days to get the most out of it as quickly as possible?

SPEAKER_01:

Yeah. But I don't think you need to like load every single data set that you've ever had in your like company's history. Like customer service interactions, for example, you know, are going to be a couple of percentage points of your total data set. I would really encourage, like, let's load in transactions, every transaction you've got on all channels. Load in your marketing engagement information and put the tag on the website to collect that on-site browsing browsing behavior. Now that's gonna get you to the vast majority of your use cases within the first one to two years. Start there, and you can move often move really fast there. And then when it comes to use cases, I would start by that value profiling that we spoke at the beginning. So let's have a look at you know what's driving your revenue from a customer standpoint. So who are the best customers? Who are your less favorable customers? How do we find more of the more favorable? How do we retain more of the more favorable? How do we nurture the others into becoming more favorable? And then I would start to focus around an operational RFM segmentation. And I would start to organize your business around that. That would be a report that I would look at every week within the leadership team. And I would start to back in your financial budgeting and targets to the RFM segmentation. So instead of just saying next year we're going to grow revenue by 10 million bucks, I'd be saying how much of our revenue is going to come from new customers versus returning customers. Of the returning customer revenue, how much of that revenue is coming from retaining the existing and eight gauge base versus reactivating the lapsed and start to be really focused on breaking down those KPIs by segments.

SPEAKER_00:

What's the secret to dissecting a good RFM segment? Because it can be overwhelming for a lot of people, right? If you if you haven't come through with it.

SPEAKER_01:

Reducing the noise. So, you know, in Lexo, we've got an enhanced RFM model that it eliminates a lot of the noise. And one of the issues with RFM is there's so many reasons why people can move between RFM segments. So what we do is we cap out any new customers. We we just set a model from day one, which might be all of your customers, and then we run that for six or 12 months. We have two segments for your new high value customers and your new low value. They've got crypt names. But then that way the new customers are not creating a source of recency but diluted lifetime value and frequency. You fix the population of existing customers in the model and we structure the RFM segments so that there's pretty clear pathways. If someone doesn't take an action, this is the segment they go to. If they take an action, this is the segment they migrate to. And we report on the movements, so how the segments are growing and where they're migrating from and to.

SPEAKER_00:

And they're all pre-built segments that once you plug your customer data in, you'll fall into an I assume you can adjust based on your business model. Absolutely.

SPEAKER_01:

But just that fixing the population of customers that are in the RFM and understanding the migration path, that makes operationally a much easier thing to report on and to hold people accountable to.

SPEAKER_00:

Beautiful. All right, David, the next 12 months for Lexa, apart from keeping Aaron out of pubs and doing his partner meetups, what are your priorities for the next 12 months?

SPEAKER_01:

We are geographically focused on North America and Asia Pacific. It's all about our AI roadmap. So for us, it's continuing to build on the capability of Lexi to help brands reduce the input effort to get to a strong understanding of their customers and automate the customer engagement so that you can go from insight to improved customer expectation, retention, and growth with less people and in a more measurable and predictable way.

SPEAKER_00:

Awesome. That's exciting. If people want to learn more about Lexa, obviously look out for the amazing merchandise at all events near you. But what are the other ways that people want to dive deeper into some of the things that we discussed today, David?

SPEAKER_01:

Look, we offer a live demo of Lexa with the customer's real data. So if you go to our website, there's a request a live demo form, you fill out a contact us form. And as part of the sales process, you can connect, for example, connect your Shopify and Clavio accounts to Lexa, and we'll show you your data live in the CDP during the sales process. And we have a use case library of 30 common use cases. You can pick eight to 10 use cases, and the team will demo how you solve those use cases live in the product with your data. So we take away all the implementation risk and you can get a real feel for what it looks like.

SPEAKER_00:

That's awesome. And I will say you've got some of the nicest and most intelligent salespeople in the industry. So that will be a pretty harmless process, I would say. It's David, thank you so much for joining us on Ad Descartes today.

SPEAKER_01:

Nathan, it's been great to be here. Thanks for having me.

SPEAKER_00:

There you go. I genuinely hope that clears it up for you. It helped me explain where a CDP fits. I've always had a pretty good understanding of it, but just as a refresher to know where CDP sits in a modern tech stack, for me that was really, really important. And when you might use a CDP rather than rely on the tools that you've already got. I really like that it's an independent source that connects all your platforms together so you've got one central source of truth. So whether you're using Lexa or another CDP that's out there, I hope that really cleared it up for you. All right, here are three ideas that I think you can take from that episode that aren't necessarily Lexa related, but you can apply in your business. Number one, identify your power 25. Run the Lexa style analysis that David mentioned, and you'll likely find the top 25% of customers drive around 65 to 70% of sales. As David said, if you're luxury, that might be higher. If you're commodity, that might be lower. But find those customers that are driving the majority of your sales and build your strategy around acquiring more of them rather than focusing your attention on every customer. Profile their buying patterns, their channels, their product journeys, and focus your budget and your attention where that audience lives. Number two, fix your match rate before it breaks your metrics. We talked a lot around identifying data and how to match. David was on the side of making sure we use multiple sources and getting certainty over those matches. I would say that if you haven't done anything in this space, any match is good at this point, especially as channels open up and we become more diverse rather than relying on the website itself. I mean, we've already seen what GPT is doing to the websites, right? So we're gonna have more diverse channels. We really need to get those match rates happening. What I especially loved was the intelligence around matching marketplace transactions from TikTok Shop, should be launching next year, and Amazon so that we don't have fragmented customer data and duplicate profiles. That is really, really important. If we can identify customers, especially those that are outside of our own channels and then remarket and retarget to them effectively, we're gonna drive our customer acquisition costs down so far and make those additional channels viable in the long term by increasing lifetime value, repeat purchase rate, and churn analysis. Make sure you have really clear matching rules in place to bring your customers together. And the last one, which was a really practical tip, was add data capture to your order confirmation page. It's not a page we probably give a lot of thought to, but David's suggestion of embedding a very, very quick survey or questionnaire into that order confirmation page of why did you buy this? Who are you buying this for? What would you like to buy next? Is really great information that we probably don't make the most of at the moment. So don't wait for a post-purchase survey that might never get even opened, let alone answered, in the email. Instead, use the thank you page to ask one short, high-intent question, such as, was this purchased for yourself or someone else? Or what was the occasion for your purchase? It's zero-party data gold collected when your customer's most engaged with you, and it's unavoidable. It's a really smart tactic. All right, if you want to discuss anything to do with customer data or anything we discussed in this episode today, jump on over to the Add to Cart community. We have a bunch of listeners in there, over 500 e-commerce professionals who are constantly discussing new platforms, new ideas, asking questions around everything in e-commerce. We'd love you to join us. It's free. You can join on over at AdDocart.com.au and we would love to see you there. As always, if you are listening to Ad To Cart podcasts on YouTube, on Spotify, on Apple, hit that subscribe button because we want to bring you episodes every week that continue to inspire and give you new ideas for your e commerce business. That's it for this week. I'll see you next week.