MarTalks- The #1 Ecommerce and MarTech application podcast
The Martalks Podcast is the leading source of the information and news from the Composable Commerce, MarTech and supply chain applications industries. Hosted by Darrell Rosenstein, the founder and managing partner of The Rosenstein Group. www.rosensteingroup.com https://www.linkedin.com/company/the-rosenstein-group
MarTalks- The #1 Ecommerce and MarTech application podcast
Martalks E-Commerce Innovation with Eli Finkelshteyn, Founder of Constructor
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Martalks E-Commerce Innovation with Eli Finkelshteyn, Founder of Constructor**
In the latest *MarTalks* podcast episode, I had the pleasure of sitting down with Eli Finkelshteyn, the founder of Constructor, to talk about how *real* innovation in e-commerce isn’t just about flashy features—it’s about leveraging technology to drive actual business outcomes. Spoiler alert: Eli and his team are doing just that, and then some.
Here’s why Constructor’s technology is one of the most wildly successful and widely adopted composable solution:
---
### **🔍 Reinventing Search with Data Science**
Eli’s journey into e-commerce started with a frustration we can all relate to: stinky search engines that don’t deliver. Legacy tools like Elasticsearch and Solr were fine for basic queries, but they couldn’t prioritize what really matters—conversions and revenue. So, Eli built Constructor to deliver just that, and oh boy does it hit the cover off the ball. By leveraging clickstream data and machine learning, Constructor optimizes search and discovery experiences to ensure shoppers find what they want *and* businesses see measurable results.
---
### **📈 Real-World Wins**
- **Sephora:** What started as a small implementation for Sephora’s U.S. iOS app quickly snowballed into a global rollout across platforms, languages, and even email recommendations. Why? Because Constructor kept winning A/B tests by delivering higher conversions and revenue per visit.
- **Recommerce Marketplace:** For a secondhand clothing platform, Constructor delivered a 5.5% lift in search conversion rates while also boosting average order value (AOV) and revenue per visit (RPV). That’s not just incremental improvement—that’s game-changing.
---
### **🤖 AI That Actually Works**
Let’s talk about Constructor’s AI Shopping Assistant. Unlike generic tools that spit out irrelevant results, this assistant handles complex queries like “What should I buy for camping with kids?” with ease. It doesn’t just answer questions; it integrates seamlessly into existing systems, giving even smaller players the ability to compete with giants like Amazon.
---
### **🧩 Composable Commerce Done Right**
Eli is a big believer in composable architectures—but he’s also realistic. He sees composable solutions as tools that should empower businesses, not overwhelm them. Constructor’s API-first approach allows companies to replace legacy systems like Endeca or Solr in record time (we’re talking as little as nine days to go live for the win!) while delivering immediate revenue impact.
---
### **🎨 Balancing Tech with Creativity**
While AI is great at automating the grunt work—like managing synonyms or correcting misspellings—Eli reminds us that human creativity is still king when it comes to merchandising strategies and brand storytelling. Think Walmart’s iconic pickup truck display; no algorithm could replicate that kind of magic.
---
### **🚀 Moving the Needle**
At its core, Constructor isn’t just about technology—it’s about the ROI. From their unique proof schedule (showing ROI before clients commit) to their 98.5% retention rate over three years, Eli and his team have built a platform that delivers trust alongside innovation.
---
If you’re in e-commerce and haven’t explored how cutting-edge tech like Constructor can transform your search and discovery experience, now’s the time to start asking questions—and getting answers that actually move the needle for your business.
💡 What do you think? Are you seeing opportunities to optimize your product discovery? Let me know in the comments!
#MarTalks #ECommerceInnovation #SearchAndDiscovery #ComposableCommerce
Rosenstein Group: martech & ecommerce executive search
Rosenstein Group is the only martech-specialist exectutive search firm. For over 20 years, we've been matching leadership talent in sales, marketing and customer success to pioneering startups in ecommerce, supply chain and sales enablement, and for digital agencies.
Our experience with recruiting e-commerce technology leaders for companies like SAP Hybris and Demandware has established us as an authoritative and trusted liaison for startups seeking their first sales leader, and scale-ups on their way to IPO.
Visit RosensteinGroup.com to find out more.
Welcome to MartiBox, the number one podcast for e-commerce and marketing applications. Martox is devoted to covering the latest technology developments that drive the global commerce ecosystem from advertising to last mile. Enjoy all of our content at Rosenstein Group.com slash MarTBox Dash Podcast.
SPEAKER_01Welcome to Mar Talks. I'm your host, Darrell Rosenstein. Martox is where we discuss contemporary issues in the composable e-commerce economy. And today we're joined by Ellie Fickelstein, the founder of Constructor, the preeminent product discovery solutions firm that's in the Mock Alliance. And Ellie is going to talk to us today about what you can find, what you can't find, why you should be looking for it, how you should be looking for it, how you will be looking for it, and uh, you know, maybe where my keys are from my uh 1972 Volvo that uh I have not seen in forever. Ellie, thank you so much for joining us. My pleasure. I hope I'm helpful with the keys. You know, uh that's that's a long shot, but you know, I thought I'd put it out there. You know, um Ellie, search is something that's always fascinated me with within e-commerce. You know, we think about the the kind of granddaddies of search, Yahoo and Alta Vista, uh, and all those guys. But I mean, I've worked with quite a few search companies in my search practice, no pun intended, you know, including Fast, you know, Autonomy, Exile, you know, Verity, you know, these are we're going way back. We're going in the way back machine here. But, you know, what on earth with with a you know, at that time when you first started working this, Endica was still the kind of heavyweight. Uh, what was it that uh possessed you to A start a startup when you were a humble data scientist? And what was it that really directed you to do a product discovery solution?
SPEAKER_02So my my background is um I was studying computational linguistics in in university. I got kind of obsessed with languages and and also data science at the same time. Um and that long story short led me into search because search, it turned out, had like every problem that you could imagine that you would want to work on as a computational linguist. Like you get everything from the really, really interesting like language-specific problems, especially if you're dealing with multiple languages. Um, but even if you're dealing with a single one, like there's still like the language understanding part of it is really interesting. And then on the other hand, and honestly, like even more importantly, you have all these really interesting data science problems that almost like bleed into psychology of like, you know, when a given person is searching for something, what do they really mean by that? How do you figure that out? You you really have to use statistics. Um and like what is the goal? Like, how do you measure success? And you've got these like really interesting metrics definition problems as well. And so just I I fell in love with the problem space. Um, I was doing it as a search engineer at both full-time at a company called Shutterstock and then afterwards uh at consulting. And I kept running into the same problem, uh, which was that every time I would be building on top of the same sort of open source solutions. Um back at that time it was primarily Elasticsearch and solar. And like these were great solutions, but they weren't specifically meant for e-commerce. They didn't natively take in this really, really valuable clickstream data that exists within e-commerce that really gives you so much signal about what somebody actually wants. You know, things like uh add-de-carts, um, concepts like inventory, margins, things like that that are important to the business. These are things that you really should be learning from. You know, if like you've got a great product that was selling really well, and then all of a sudden, like the most popular sizes run out, what do you do? These are very, very specific e-commerce things. And so for us, for me, it was like I can either keep trying to hack that into these like one size fits all general solutions, or we can go and build something that really takes this stuff in as a first class citizen. And if we do that, like my bet, and we've been lucky enough that it's turned out well so far, is uh was that we could just do a much better job in terms of getting the e-commerce business metrics that the sorts of companies that we work with, the Sephores and the Underarmers of the world, um, and just be able to do that much better than anybody else in a way that was provable where they could A-B test it. See, they're making so much more revenue with us than with someone else.
SPEAKER_01Well, I'm Claire, you're you're you're trying to figure it out because I'll tell you what, well, uh, I'm still having issues when I want to get my 1980s winter freak on and do a search for a chartreuse muffler. I mean, you wouldn't believe the results I get. I it it's like the the search engines are having a fight to figure out what what the most ridiculous representation of that is. Uh needless to say, uh, it isn't what I've seen as yet. But any of you out there that are listening that do merchandise, a chartreuse muffler, you'll have my email at the end of this.
SPEAKER_02I'm gonna have to go test that search afterwards. I don't know anybody that sells a chartreuse muffler, but don't I'm joining it.
SPEAKER_01There's there's companies that are gonna pr they're gonna insist that they do, and they're going to be sending me a coupon to get two for one. I can guarantee it. I can guarantee it. And free shipping. Free shipping, yes. Well, it's obviously clear in in reading the press uh about constructor, which has been amazingly positive. I mean, you've got some raving fans, you've got some raving fans uh that uh you you are relentlessly focused on increasing the revenue for your clients. In fact, I know that the latest IDC knowledge discovery report had constructor as a leader, which is quite an accomplishment. That is a great analyst organization. They really do look for results, they interview those customers thoroughly. So well done there. But well, that must have been great. Uh certainly an ego boost for you and the team, a couple of chest bumps in the office. I I also look up some of the metrics that you guys have delivered in your case studies, and some of those were pretty darn impressive. Kind of some awe shocks. Wow. Uh, can you share how you felt when you found out that you know Petco was getting a 13% bump in conversion rates or you know, Princess Auto? I don't know what exactly Princess Auto is, but they saw a 22% bump in conversions.
SPEAKER_02Princess Auto is a really great retailer in in Canada. If you happen to be in Canada, you should check it out. They're wonderful, um, really wonderful people who work there also. But um, yeah, it felt really good. Like anytime that we can we can tell our customers that we're going to deliver something for them, and then we get to beat the amount that we told them that we were gonna deliver, like that always feels really, really good. It feels like I can get excited about what we're what I'm doing when I wake up in the morning. Um, it also just means that like we had this hypothesis of what we would be able to do, and seeing that actually work out also just it makes me really, really happy. And um, for as long as we're doing this company, I'm gonna keep trying to do that exact same thing for everybody that's willing to put their trust in me and willing to do business with us.
SPEAKER_01Well, your scientist was showing for a second there. You hadn't high you had a hypothesis, and uh you you you actually answered, I I appreciate that. I don't think we've had a hypothesis yet in this show, so thank you for giving away. I hypothesize that this will be a good episode. We're just gonna go there and I hope I didn't jinx it. But uh, you know, uh one of the other things that I always look for in a firm is what are the customers saying? I mean, I definitely before I engage with a client, um, I'm looking at, you know, what's their reputation score? What do they uh, you know, what is their CSAT rating if they have one? And you guys definitely do a fabulous job of making your clients not just happy, but as we've discussed, realize real results, you know, a serious ROI. And customer success is something you get, you know, it's the linchpin is to show users how that solution is impacting or would impact their business. Could you share a little bit about how constructors proof schedule process allows prospective clients to see that ROI using their own data and walk us through how that process works?
SPEAKER_02Yeah, sure. So we we have a general process um for trying to ensure customer success, like the customer actually feels that they were successful with constructor from the first time that they start talking to us all the way to when they're launched and they're live and they've been with us even for you know three, four, five, six years. Um what we want to start with is aligning on what success looks like to them. That's at least to me the most important thing. Um, you know, if if like I think that the most important thing is, let's say, revenue, but for them, like they're not actually trying to focus on revenue, they're trying to maybe focus on like profit or maybe they're trying to focus on conversions or something like that. Um, it doesn't matter what I think, it it matters what what they think. And so having that conversation with them initially of like, what is that metric or that set of metrics that um you're gonna define success by, and how big of a lift do you need to see in them in order for for you to feel successful if you install constructor? And that's the very first step because if we can't align on that, then like we're gonna go and they're gonna install it, they're gonna hope for one thing, maybe we deliver that, or maybe we deliver something else. We don't really know what what they want. And so like I think that we would just wind up butting heads in that case, and we don't want to have that. We want to set the customer up as much for success as possible. The the next step is trying to figure out a lightweight way of um showing the customer that we're capable of doing that. And it's also showing ourselves, honestly, too, because you know, if the customer says, like, I need a 100% increase in in um revenue, like it's unlikely that installing a new search engine is gonna get you that. Um and it's better for us to tell them, like, that's probably not gonna, I mean, maybe like if we get lucky, but it's probably not gonna happen. Um, better for them to hear that before they install it than afterwards. So the proof schedule that you mentioned is is this process for figuring out a in a lightweight way without them expending a bunch of time because they've got a bunch of different priorities that they could be uh putting um time into, figuring out how much can we actually drive for them. So the way that it works is you send us a product catalog and you install two lines of JavaScript onto your website, which starts to collect your click string data. So what people are clicking on, what they're adding to cart, what they're purchasing, what they're scrolling past, et cetera, et cetera. And what we can do using that is we can start to look for places where we see holes, where we see gaps, um, places where maybe somebody is having a subpar experience where, you know, like uh I remember there's an example of um somebody needing to uh search for Air Force ones, and it would work well with the number one, um, but it would not work well with it spelled O-N-E-S, or maybe it was the other way around, I forget which one. But you could see like one of those searches had this like really great conversion rate, and the other one had like a really bad conversion rate, and only some of the people reformulated. And so you look for examples like that, and you add all of them up where people could have a better experience if they had something like constructor, and then you come with a business case and you say, okay, like here's what we think we could do for you. Here's what it's based off of, which is all these examples that we're finding on your site where we think that we could do better. We'll show you those as well. And then we say, you know, do you is that exciting enough? And if it is, then then great, then we should move on to the next stage. And if not, like that's okay. Um, we we both learned something early before the customer needed to invest a bunch of time. And so in that case, we wouldn't be able to work with them, which is unfortunate, but hopefully they'll at least tell people still that they talk to, like, look, you know, constructor, though those folks are not just gonna sell you something that they don't that you don't need. Um it it also bakes into one of the things that I'm proud of the company is our our retention rate, which over the last three years averaged out at 98.5%. So very, very few people that um that start using constructor will not use constructor anymore. And I think a lot of that is just ensuring like only the people where we can actually get success for them wind up using it. So that's the proof schedule. And then finally, after all of this, um, if the company decides to to work with us, which I hope that they do, um, there's always a question of what have you done for me lately, which is a very fair question to ask. So it's letting them, if they want to prove out, do we actually get them the revenue lift that the proof schedule showed? Maybe they want to run an A-B test, that's okay. Um, maybe they want to do pre-post analysis. If they don't have the time or the expertise to run an A-B test, that's also okay. But then after they have it installed, like what kind of variations of our algorithm can we test for them as we're doing additional development, as we're improving to be able to prove for them that constructor in 2025 is doing more for them than constructor was in 2024. And now was doing more than what constructor was doing in 2023. So we don't need to only prove to them that we're the best solution at the time that they decide to use us, but that it's getting better and better and better over time and we're constantly improving and we're constantly giving them more bang for the buck.
SPEAKER_01Wow. Upgrading service just automatically. Yeah, I, you know, that's that's what initially attracted me to cloud native solutions. I mean, having having an AP API-led cloud native microservice architecture solution, you you not only can deliver uh a great experience out of the box, you can continually improve it with no disruption. There's no disruption. Just like, by the way, did you know that your slow-moving inventory has increased its flow-through rate on your site by 15% over the last year because we're it's showing up more. People are seeing it more than they weren't before. I mean, that's that's that's a great result. And customers need to be made aware of that. Did you know how great we are? It's like it's it, you know, uh it's not bragging. It's like, do you know how great you're doing? Everybody needs that. Who wouldn't want to tell that to their boss? You know, that uh search uh that product discovery solution I got, uh constructor, yeah, it just increased our conversion rate another five percent this quarter. It's like I I'd like to bring that up.
SPEAKER_02I think it's a meta one. It's um, yeah. Like if I'm putting myself into the customer's shoes, like if I was there, I would want to make sure that I've got the best solution every single year. And if they make that job easier for me by telling me what they've done for me lately, then it makes my decision easier. So I'm trying to do the same for our customers.
SPEAKER_01Yeah, I I think about the world of retail in general, and where you've got CEOs that are former merchants in most cases, they they're not e-commerce veterans. And they're they're just looking at e-commerce as a store, and they're not appreciating the fact that these investments are paying significantly more dividends than say remerchandising a point, their their flagship point on Fifth Avenue. It's uh it's a it's just legions of difference, but you know, you you gotta make the client aware of that. Uh and obviously constructors doing a great job of uh turning those those bold CIOs and CTOs that have chosen to go the composable route into heroes and make sure they have a story, their heroic journey to share with their peers. But let's let's uh divert here a little bit. This is this is something that uh has been very much on on everyone's minds is you know, the rise of AI and AI-powered shopping assistance. I mean, perplexity being one, perplexity now having a buy now button if you do a search with it, uh, you know, enabling you to purchase directly from your results. Uh Amazon and Walmart weading into that as well with their search engines. How is constructor optimized to integrate and complement these AI search agents? And are are you bullish on that new technology?
SPEAKER_02Yeah, long story short, yes. Um I think I think it's important to have a balance here. Like I'm I'm simultaneously bullish on the technology. Um, and I I try to have recognition that we're still in the Wild West and the way that things look right now and these initial shots that people are taking at it, this might not be exactly how things evolve over the next five years. So I'm bullish that things are going to change and we're trying to do a lot of experimentation internally. So like we, for example, we have our own AI shopping assistant. Um, we wanted to make sure as soon as that technology came, became available, um, that it wasn't just something that our customers needed to look over at like the Amazons and the Walmarts of the world and say, like, oh, that's nice that they have that. I wish I had that. But we wanted to make it easy for them uh to be able to do that as well. And the nice thing is uh when customers integrate with us, we already have their product catalog. Uh we already have their clickstream because that happened by the approved schedule we were talking about before. So if they wanted to try something new like that, um, there's a very, very low commitment. There's not a whole lot of integration that needs to be done. It's basically just the UI because we already have everything that we need in terms of data to build the algorithms ourselves. It's just, you know, making sure that they have a partner. Hopefully, we can be that partner that is on the cutting edge of this stuff that's willing to take those bets along with them. Um But then there's this also a question of like how do things evolve? And so there we don't want to get married to just one uh idea of how this can work. So, like, for example, our AI shopping assistant, you can use it um to give long form queries. Um, so you can you know go into it and say, like, I'm going camping with my kids for the first time, you know, they're four and seven. What do you recommend that I buy? And it'll actually understand that. But then also, like, can it generate um questions and answers on a PDP? You know, what are the typical questions that somebody might ask here, which we can figure out from reviews? And what are the typical, uh, what are the best answers for those questions? Which we can figure out largely from like the descriptions, maybe there's something in other reviews, maybe there's other data that we can get this stuff from. It's a different kind of discovery. Which one of those becomes more important in the future? Like that's the thing that people are still testing. You can see the Walmarts and the Amazons and the perplexities of the world testing this stuff. We just want to make sure that the tools that they have, our customers have access to the same or better.
SPEAKER_01I I think about the evolution in this space. I mean, I remember first talking about the idea of clientele uh with Rob Neibauer shoot 20 years ago. And it was going to be a POS-based system that uh the shopping clerk was going to say, would you like socks with that? Uh, essentially, if you were buying shoes or something to that effect. And now here we have AI essentially serving as your concierge and going directly to the brand to make a purchase, which um has some marketers' concern because it essentially will obviate the need for much of the Martech stack that has been built around the customer surveillance model of you know click click streaks, where you're looking at every individual choice that a customer makes and trying to calculate down to the nth penny um what ad drew people to the site because the agent's going to be impervious to that. And the agent is going to learn what your preferences are, what your size are, what mood you're in, depending on how you ask a question, and go directly to that brand to make that purchase. It's it is the Wild West. It is the Wild West. I mean, so I love the fact that you know you're thinking ahead and going, all right, we don't know how this is gonna work out. We just uh got to be flexible. We got to make sure that our customers have the best assistant that they can have in this brand.
SPEAKER_02Did the agent think what it's worth, like it's actually a good example because that that agent also has click screen, like it also is gonna be going and just looks different from what a human has, but you know, it's also going into the website and like it's looking for something and maybe it's trying out different searches and things like that, and it's giving you indication about what it likes and what it doesn't like. The big question in my head is like, do shoppers want an agent or do they want to evaluate these things themselves before before buying? And on that question, I haven't really heard at least right now, maybe it'll change in the future, but I haven't heard very many people being like, I want an agent to buy me a new pair of pants sight on the scene. It's just it's a lot of trust that you're putting into a thing that way. Maybe we get there eventually. We want to be prepared. Prepared just in case it happens. But I think that brands that want to own that channel, that that direct relationship to their customer, there's still a good reason to believe that that's going to be possible in the future and they're not going to just have to talk to an agent all the time.
SPEAKER_01But we'll see. I mean, there's going to be shoppers and there's going to be people who don't want to shop. It's always the case. You know, maybe the agents will actually make it easier for it easy enough for people that weren't shoppers to actually look at choices that now they do want to look at choices. And conversely, folks that did enjoy going out shopping, it's like, oh wow, I really can save a lot of time by having my agent do this. Well, it's it's it's exciting. It's exciting. Well, I've I I used that word uh earlier, composable. Um, and it is, in fact, the elephant in the room. Uh, I'm a huge fan of composable architectures because they're future, they allow a merchant to future-proof their stack by building it all out of the components that work best for them, the best content management system, product discovery, uh, the best personalization engine, the best commerce platform, OMS, take your pick. Uh, and you know, they're all API-led, cloud native, uh, microservice architectures. So if a better one comes out, if somebody isn't as committed to continually improving their experience as another vendor in that category, they can swap it out quickly, easily, a couple of weeks, a couple of months, tops. So composable architectures have obviously been gaining traction in e-commerce, but much faster in Europe than in the United States. Here in the United States, I would have to say adoption has been slow. Why do you think that is? And and you know, how does constructor's composable features help clients transition more easily to that model?
SPEAKER_02Yeah, so I I think part of it comes down to the definition of composable.
SPEAKER_01Um I knew you were gonna say that. I knew it.
SPEAKER_02I think when when you look at like a lot of the really successful companies right now, many wouldn't describe them as composable as a company, but then they also have composable solutions like Shopify and Salesforce are great examples of that. Um on the flip side, you know, there are like pure composable companies that are also fantastic. Uh I think that composable it you like with anything, like it's it's not great to treat like with any technology at least, it's it's it's not great to treat it as like a religion where it's like, you know, you have to have it or you don't have it. It's a tool. Um we have great partners and that are would traditionally be described as composable. We have great partners that would traditionally be described not as composable. Um for us, we we treat, we think of it as a tool internally for what constructor is. Um for product discovery, I personally think it makes a lot of sense because product discovery is something that you would fit into an existing stack that you already have. Um you know, you you can install it as part of a new one, but like a lot of the time when we're coming in, it's to fit as uh fit into an existing stack that they already have. And what they're looking for is they're looking for a little bit more revenue out of the existing product discovery platforms that they have. They want to swap that out and get more revenue. Um for that, I think composable is the very best thing because you can swap us out. Uh hopefully we can do a really great job for you. But theoretically, if we can't, like you kind of have like you know the gun to our head of like either do this or it's composable, so you can very easily swap something else out. And I think that that's a good thing. Like I think that puts us into or any vendor, it puts them into the right mindset, which is like you have to earn this business every single day that you're working with them. And if you don't, then it's easy for them to leave. You're not gonna be able to stick around just because you're sticky or because it's difficult to pull out or because you're a monolith. You you have to earn the business. I think that's better for the customer. And I think it's more it's better for us as well because it gives longevity to the company. Like we we don't want to stick around just because the customer has to keep us. We want to stick around because we're doing a good job for them. We want them to tell us if we're not doing a good job for them, because we should fix that immediately.
SPEAKER_01Yeah, yeah, I definitely don't want to have a customer just because they have lack of options. I mean, in my business, they definitely have plenty, which is so it's not an issue, but uh I I think I feel a little guilt. But when we look at the adoption of constructor in particular, you know, you've had multi-billion dollar clients, they implement constructor not quite in a matter of weeks. I think it I think you do, it does take uh a month, but it I think you might have uh a couple of examples where it's it's been weeks, and then they realize a five, ten, fifteen percent uh jump in conversions.
SPEAKER_02That's probably it's fun actually just now that you're mentioning our record is currently nine days. I'm I'm saying because we had a recent one that did it in that period of time. Nine days. The previous record was 10, and so we were like, can they beat it in the yeah? Anyways, neither here nor there. It's fine if they take longer, also, but that one just made me really happy.
SPEAKER_01But it's just, I mean, it's mind-blowingly easy comparatively. Companies think of the word replatform anything, and it's it's just like they they run for cover, uh, and they think about the the their capital that they're gonna have to spend to get a project online. And now you've got now we've got an entire suite of solutions to make the the their commerce uh stack faster uh and more productive and more profitable than it's ever been, and it's easier than it's ever been to make that transition. I think that you know it's having an experience with constructor probably makes a client more likely to pursue the rest of the stack to improve their uh their digital experience for their customers than it was ever possible because they can pick it, they can strangle the fig. Uh, which again, when when I think about composable, that's I'm thinking about that API-led, cloud native, microservice architecture. Uh, and that's kind of what that I'm I'm sticking to that. There's companies that have decoupled some of their functionalities, but it's still pretty big and furry. But uh, you know, you got constructor replacing mega applications like NDECA or solar uh again in a week. Especially it in a week. Like, wow. Okay.
SPEAKER_02But you mentioned I I think just to like what you were saying before is is also just absolutely fair. Like when we're replacing something within a week, very often that's enabled by the technology that they already have. Like they already have like a great composable piece of software, and you know, like you you can do that sort of stuff with some of the companies I mentioned beforehand. You can also do that with like the commerce tools and the big commerces of the world, and those folks are fantastic. It's composable to me, it's it's it's a tool. It's a tool. It's using it where you need it. And a lot of the time that's what enables you to be able to install these things so quickly.
SPEAKER_01Yeah. Well, I mean, installing something that big with that much of an impact in nine days is huge. Can you share a another kind of of your favorite success story where constructor, you know,'s discovery platform significantly improved a customer's business metrics or or or their customer experience?
SPEAKER_02There are a few. Um at this point, there are a few. We've been through something like 80. No, sorry, that's all. It's more like 90 A B tests at this point, um, with companies that they were installing. So, like I'll give I'll give you two. Uh one of them I can I can mention the name of the company, the other one had just finished. It's just fresh, and I don't think I've got marking right, so I can't mention the name of the company. But um, the the first one is uh one of our first customers was a company called Sephora. Um, and those folks are fantastic. We've had the pleasure of working with them, I think, for something like five, six years now, or something like that. Um, but they uh they've got a really strong testing culture, and so they they started off by installing just a very small piece of constructor. They um they said, we'll give you a shot um after we go into the proof schedule, but that shot was only on their iOS app and only in the United States and only for search. So they weren't using our browser or our recommendations or any of our other technology at first. And we won that A-B test. Um, and then afterwards they said, okay, cool, like you've proved it out. Um now we want to test another part of you. Like, we know that you've been doing this on iOS, can you do it on desktop and Android as well? And so they did that test and we won that as well. And they're like, okay, cool, we'll install it there too. Can you do other languages? Um, we've got a French site in uh for for French for Canada, French Canada. Um, so they tested us for French as well. And we did that allveyeu as too. I'm sorry? Melveilleux. Exactly. I don't know what that means, but hopefully they don't think um but any so we they did that test, and then eventually they installed us for recommendations the same way, and they started using us for email recommendations even. Um, but with each of these things, it was always after a test. And so that was just an incredibly fulfilling and like nice experience for us where like it was hard work, right? We needed to do all these different tests, the tests were not zero work. Um, but it really proved out that every single part of the platform individually, like it wasn't just like a you know mixed thing, it was every single part of the platform individually was able to earn its place. That made me really happy. There was um another one um that that just finished recently. This one I don't remember if I can use the name, but uh, this is a company where they're a two-setted marketplace. Um they they would sell uh used clothing. And so there they had these doubts initially about us. They were like, you know, can can the clickstream approach work even when you only have one of every single item that you sell? Right? Because like the you know, sweater or whatever that that Ellie is wearing right now, like there's only one of this of this sweater, even if it's made by by whatever company, like there's only one in this quantity, this quality that I'm I'm selling right now, and so it's difficult to compare that to other ones. Um they they also set a pretty high benchmark for it. They said, you know, we're only going to be able to declare success on this thing if um we get to at least a 5.5% increase in search conversion rate, specifically only on search. Uh and the the technology that they had, they had done quite a bit of work on it. It was um an in-house team that had built on top of Elasticsearch, but then they A B tested it and we surpassed um their benchmark, but then we also added additional uh AOV on top of it. And then we also added additional RPV on the browse side, and so it was just this like knock out of the parts success from when their CFO was looking at it afterwards, and they went to 100% with us much faster than they expected to. And um that just that felt really, really good. So we were all high-fiving internally about that as well. Outstanding. So that was in recall, that was on a re-commerce site. That was on a re-commerce site, exactly.
SPEAKER_01Wow. That you know, that's that's a fascinating issue because again, you go to a company, uh you go to a site, and I'm I'm an avid uh real, real uh shopper. Just uh, you know, I I I go back with some of the founders of the firm, but I think that the ability to do a search where you're looking for a very specific piece of whatever and be served up results of other things that are approximate enough that you're increasing your AOV with that, that's pretty spectacular. That's pretty spectacular. I just because you're going for a very specific thing. It's like, I want that Birkin bag. I want that. Well, I mean, I just I don't have any Birkin bags. If I did have uh purses, I would want a Birkin bag, even though they are three or four thousand dollars on the secondarian market because Air May just does a fantastic job, as does Sephora. I mean, Sephora's got a fantastic reputation. That's a very progressive technology experimenting uh organization. Anybody who I've come in contact with from that organization is uh a rabid technologist and about taking risks. And I think that more retailers should pay attention to what they've done because they could save a lot of time and effort if they watched and listened.
SPEAKER_02I think so. I'm a big fan of those folks. A lot of very good, very people over there.
SPEAKER_01I like the anthropology team too. Those guys are spot on, they try out everything and they know how to figure it out pretty quick. Um Well, we've talked about some of the AI that constructors linked into since uh its inception. You know, do you see replacing some of the things that uh humans would traditional do? I mean, for instance, merchandising, for example, does it still have a place in the AI future?
SPEAKER_02So uh at least my opinion on this is absolutely, because I think that there's this balance uh within commerce in general of art and science. On the one hand, there's there's a science where you're looking at that click screen data, and for for a human to do all those calculations when a computer can do them much, much faster, it doesn't make a whole lot of sense. And so like the things that you get out of it, like on the old world, for example, um, you would have merchandisers. And this this was not never their job originally, but people would have them going in and like putting in like synonyms um because the cis like they were using these old keyword-based um search engines like Ndeca that you had mentioned or solar elastic search, um, or they'd be like manually correcting misspellings or something like that, because those things were built as a one size fits all, and like maybe they didn't have a concept of how people would typically misspell. That sort of stuff, like that menial work, which never should have been part of merchandising in the first place. Like, I don't think humans ever should have been doing that in the first place. And I think it's a brighter world if they don't have to. On the flip side, you have the art of it. So like I remember I was I was reading um this tells you how much of a nerd I am. Uh I was reading the the autobiography. I'm sorry, guilty. There you go. Um I was reading the autobiography of um of Sam Walton, who who founded Walmart. And he had this example in there where uh they this is like in the early days of Walmart, but they took uh they needed to sell a lot of um a lot of car wash, uh like car wash uh fluids and soap and stuff like that. And so they took a pickup truck and they drove it into the store and they filled it up with these things and like it made this like really, really great display. And that's not something that I would ever trust AI to come up with. Like that's a very human thing to come up with. That's more of like the art part of it. And it was a great idea. Like you read you know the autobiography, and it's like that stuff actually sold out much faster than if it had been on a shelf. That's the sort of stuff where I want to enable merchandisers to have time to do the art, the creativity, um, the stress, the more strategic stuff. And the only way that they'll have time to do it is if we can take the menial labor off of their plates. So that's kind of how I see the division.
SPEAKER_01That's very reassuring, you know, for the artists out there, for the creators out there that uh that right now AI is good at replicating and automating very, very menial tasks, and we should lean into that and you know, spend more time doing the things that bring us joy, which is being creators, being being innovators, and uh bringing uh innovation to our stores.
SPEAKER_02Well, it means that bring us joy, but I think it's also important, like I don't know. I if you imagine a world where like everything that you read is written by ChatGPT, like that's a really sad world. Right? Like I don't think anybody wants that world. And so it's the same thing with commerce. Like if every commerce uh e-commerce site looks exactly the same because it's all power, like it's all purely AI and it doesn't have any human-driven merchandising in there, like that's also that's not gonna be a whole lot of fun. I don't think anybody wants that. So I don't think that we're gonna get there. I think the place that we get to, which is a much better place, is let AI do the menial stuff, the stuff that AI is gonna be better at, and let humans do a strategic, the artistic, the creative, the creative stuff that really just helps define the brand, helps define the retailer who they are.
SPEAKER_01Well, I'm I'm hoping that there isn't an AI-based podcast out there that's going to listen to this and go, well, you know, I think I could write some great copy. Uh that would that that would scare me. That would scare me. Well, let's let's move on to something that you and I share, uh, which is customers uh you're you're in a sales cycle with a customer, they get a competing bid from a competitor that's dramatically lower uh than theirs. And you know, you you've got to then lean back into your ROI story. I for you know, me as a as uh an executive recruiter, I talk about things like uh tenure. My tenures tend to be 40 to 60 percent longer for the people that I place than people that are hired in that role or uh time to productivity. Um folks tend to transact deals um uh 30 to 60 percent faster than folks that are hired traditionally. I mean, you've you've gotta lean into your own ROI story on that. How do you, when a customer is looking at a competing bid, convey your value to clients who might initially focus on cost rather than ROI?
SPEAKER_02I think it's I mean it's a hard question. Um, but I think there's there's two parts to it. So one of them is we need to, if if I sound like a broken record with anything, it's with with the word empathy just because I think it's just so important in general. But like part of it is, you know, we need to try to get onto the same side of the table as as the customer. And if I'm putting myself into the customer's shoes, if somebody came to me and they said, you know, let's say I'm like the CIO or something like that, and they said, you know, you can either get one search solution and you're gonna pay, let's say, you know, uh I'm just gonna use like a dumb number, but like let's say a dollar for it and you get, you know, ten dollars back. Like, is is that a good deal? Like, well, yeah, of course. Like, I'd love to trade a dollar for ten dollars. Why would I possibly not want to do that? Um, let's say there's another one that you could get where you trade $2 and you get $20 back or $30 back. Um, is one of those a better trait? I'd be like, well, yeah, like I want to try to get scale. Um, if I could trade a dollar for $10 or $2 for $20 or for $30, whatever it is, like anytime that I can get a better return on a dollar than a dollar, like that's a worthwhile thing for me to do. And so when you're looking at our technology, it's like there are things that we can do where they're more expensive to train. For example, let's say like a transformer, which I won't go too too much into, but it's what the T and ChatGBT stands for. Um it's this really powerful thing for helping understand uh what people mean when when when they're searching. But it's also more expensive to train than the older generation of things people would have used for vector search. So it's like, okay, like is that dollar that's spent on a transformer worth it on top of whatever? It's like, well, if it's giving you much more than a dollar back, then then probably it is. And so this is the thing that we really need to help our customers understand from their perspective. Like, I get it that it's it's hard. Like everybody is trying to tell them, like, oh, I'm gonna make you so much more money, and how do you trust one of these versus the other? And so that gets back to the proof method for the dollars that we were talking about before. I don't think that you should trust like me or you know, anybody within our space just by the words that they're saying. Um transparently, like that's um that's better for me because I know I'm not a very good salesperson. Uh so if it's just trusting, you know, whatever that like somebody's gonna be able to outsell me. Um I'm just like make them prove it. Like there's there is a proof methodology for this, whether it's a proof schedule or something else, like you can decide on it. But but make them prove it. Like if they can prove to you that that's the trade that you're making, then then great, then you probably want to make that trade. Like you want to try to trade every single dollar that you can for more than one dollar because that's always a worthwhile trade. We just need to prove to you that that's the case. That's the first half of it. The second half of it is um and this is a decision that like I've had uh investors previously become upset at me about, but but it I want to I think it's the right one. I want to go for as long as possible. Is our pricing is more straightforward uh than than our competitors. So if you look at it, it's just a flat rate. Like what we do after we give you after you go through the proof schedule is we have enough data to basically be able to say, like, here is your price, and regardless of what happens to your traffic, regardless of what happens to your product catalog, anything like that, like for the life of the contract, this is what you're gonna pay. Um you know, don't don't do anything crazy, please. Like, don't just start sending like random traffic from like another site or something, but like this is it. And because of that, like we don't charge customers more than they expect. It's that hard conversation up front, but it's not a hard conversation every month when they're getting some additional bill they weren't. expecting. And when you compare that to other folks, like we we have had a number of customers, for example, come over from another competitor whose name I won't mention, but uh part of the reason was it looked cheap. And then they realized that they were also getting charged, they were getting charged per query and that bot traffic was included in that. And they had all of this bot traffic that they weren't expecting at first that all of a sudden brought up the expected traffic, the decimation of the contract was initially based on way more. So they wound up paying two, three times as much. And they were like, okay, now I don't want to be here anymore. The reason I started doing this at first was to save costs and that actually saving cost. The unfortunate thing for us is like that sort of thing will oftentimes win an initial deal against us. But the nice thing is that a lot of those customers afterwards they'll they'll come back and it is painful to lose those deals eventually, but like I think that it's still right to be transparent with the pricing. I think it's better to do it that way. And I think that's the thing that enables us to have a better retention rate and like eventually it'll it'll build on itself so I'm still happy.
SPEAKER_01Well I think that any you know like a constructor is opex versus capex as well. Yeah. So I mean there you are. You're you're essentially uh retiring a capital debt with the solution.
SPEAKER_02That's what we're trying to do. Um I know we're a little bit over here unfortunately I'm gonna need to run in in in a second here. Maybe one last question if it's all right.
SPEAKER_01Yeah well as someone who spent 25 years helping startups find the right talent you know I'm always fascinated by that team building process. And you know what were your biggest challenges when you were hiring your first executive team and you know what what qualities have proven out for you as being just essential and putting people that have been successful with you head and shoulders over the folks that haven't um the the initial challenges there were a lot.
SPEAKER_02I mean there's there still are but if I needed to if I needed to put it into two buckets and say one of them is is hiring for uh people that are going to be filling a role that I've never done myself. So if I'm trying to interview for for an engineer like for that stuff we we had a really good track right at the very beginning of the company like because I had I had done it before um so I had an unfair advantage. But then if it's like going and hiring like our first head of marketing or our first head of sales those those were much more challenging to do because it's me trying to evaluate something that I don't understand nearly as well as the person that I'm talking to and that's true for pretty much every single person that I'd be interviewing. So that was that was half of it. And the other half the the uh main bucket was being willing to disagree with people where an incredibly incredibly smart people um that had done it before but where their advice just didn't match the sort of company that we were trying to build. So like we had many smart people that would give us advice you know investors that would give us advice who I very respect a whole ton who had done it before. And so like on paper like their opinion should should matter much more than mine. But they weren't trying to build the same company that we're trying to build and I think every company is is unique and this you know the thing that works at one company might not work at another one. And realizing that and then also being willing to disagree with them and say thank you very much for the advice like we still really appreciate it or I still definitely want to hear it. But if there's a specific reason for why that advice might not work out for constructor specifically being willing to say to stand up and say like thank you for the advice I'm gonna do something else. Those are the things that made it difficult. Um the things that that I think have stood out as being some of the most important attributes of of people that work here. And again this is just for us it doesn't mean that this is the right thing for everybody I mean for every company but we really like people that are curious. Curious people tend to figure things out we we like people that figure things out. We we have a really good track record. And this this sounds obvious but I think with most comp like with a lot of companies it's maybe not as high on the list as as as for us is we want to hire people that are genuinely good people. Maybe that that salesperson that you hire that isn't willing to like hoodwink a customer like they might not make as much revenue for the company but that's a trade-off that we're that we're willing to make like we would prefer to have the good person because it just lets everybody it makes it so much easier for everybody else within the company to also be a good person. When they know that people are going to be good to them it enables them to bring out the best parts of themselves. So that's turned out to be really good for us also. Introspective people where the way that when they're looking at like who they are and where they are within their careers matches the way that we see them where that that match exists there's just a much higher probability the thing is going to work out. Like if you hire somebody that's amazing, you know, let's say they're like here but they see themselves as even higher than that, um it's gonna be difficult to come to grips on like what that person's role is going to be and what it should be and how fast they get that promotion and things like that. And you're kind of you're setting yourself up for failure. Because that person might be really smart but they might not be a good fit here. They might try to bite off more than you think that they can chew um or you know they they might uh just not be happy with the position and the responsibility that they have and so finding people where your view of them is aligned with their introspective view of themselves. I think empathy that we talked about before like that's incredibly important that the person is constantly putting themselves into the shoes of the people that they're interacting with. It's especially important with customers like you have to do that with customers otherwise you're not gonna you're not gonna get success with them and they're not gonna be happy with what we're doing for them. But you also have to do that with coworkers. You have to do that with the people that you're working alongside with other vendors that we use. And then finally grit grit and hard work like I just think there's there's no substitute for grit and hard work especially with a rapidly growing company. I don't know if startup is still the right word for us. Probably it is but um there's just no substitute for hard work and either either you're willing to do it or you're not and we want to hire the people that are well kudos to you for you know not not I would I wouldn't term it standing up to your investors when I I think our investors are wonderful. I don't want to give the wrong idea or anything like that.
SPEAKER_01Yeah I I'm I'm not I'm not uh suggesting that at all what I what I was gonna say is that they realize that they don't know it all they can offer you their advice but the reason they invest in you in the first place is because they believe in your vision and you know you being true to that vision initially when it when it counter when it contradicts some of their advice is awkward. It's awkward at the beginning. It's not an easy thing to say I I'm gonna go a different direction than what you who have been out in the world and done things similar to what I'm doing right now are telling me to do. But it's it's that fidelity to your vision that excites them that they are they're they they love it when they see it uh well obviously when it works um but it's also what your team is is really is it's what your team really rallies around. It's like wow our CEO is in charge and and they have a vision and they've they're maintaining uh uh their fidelity to that and to us uh and to who we are um but I I I have to ask you one more question before we go because it just naturally comes into that I mean leadership plays a huge role in you know recruitment and culture and you've got a leadership style uh Ellie and uh it's influenced constructors' ability to attract and retain uh some great people you know what's your secret sauce man I don't know and you could say empathy one more time I'm just gonna let you say it one more time and then we're retiring it sure one one last time and empathy I do think is incredibly important with with leadership and just about everything that we do.
SPEAKER_02I I do unfortunately if I if I needed to go quickly like I think um humility about the things that I don't know like I'm gonna need to make decisions on them anyways but being humble enough to know that I don't know the the right answers always um and and and being honest about that with people also like one it's a hypothesis it's not something that I know for sure um I think it's important to to lead from the front that's and I don't know that everybody would agree with that. I think there are lots of great great leaders that that don't do that and that works for them. At least for me like I think that it's important for my leadership style to not just tell people to do something that I wouldn't be willing to do, but to be willing to do it myself also. Like maybe not every time maybe I can't do that quite as much anymore because my time is is a little bit more um pulled in all directions. But like if I want somebody to do something and I'm not willing to do it myself and I've never done it myself like it's just I don't know that that feels kind of crummy for for me to tell somebody to do it anyways.
SPEAKER_00Thanks for listening to Mar Talks the number one podcast for e-commerce and marketing applications. Be sure to subscribe wherever you listen to podcasts and while you're at it leave a rating and review. To find out more about how the Rosenstein Group can help you fund the right leaders for your client development teams in Martech and e-commerce please visit our website at Rosenstein Group dot com