He Took ironSource to $11B. Here’s His $180M AI System for Scaling DTC| Omer Kaplan, Zyg

0:00Loukas:My guest today has spent his whole career around one idea. You don't guess what works, you test it and then you pour fuel on the fire. He built and scaled the mobile business IronSource, which essentially is the company behind app monetization and distribution tech. And he took it public for$11 billion. They tested and scaled thousands of mobile games and he's since gone on to a new venture where he's raised over$180 million, applying the same model to D2C and it's called Zig. I'm super excited to welcome to the podcast today, Omar. Let's get into it.

0:39Loukas:Welcome back to another episode of DTC Diaries. This is one that I am super excited. Omar, welcome to the show. Thank you very much. It's really great to be here. Cool. Well, yeah, I think what I find just super interesting about your story is how you've taken and a model in, I guess, like a comparable industry, but very different in many ways and applied it to DTC and we'll definitely be getting onto that shortly. But where I love to start is just give me the top-level origin story. I think taking a company public at$11 billion valuation, that's definitely not the start of the story, and I'm sure there were many failures and learnings along the way.

1:22Loukas:So talk me through, you know, in a short way, what that story looked like. So it was quite a journey. We were eight co-founders, which is really interesting because we are also eight co-founders in ZIG. And similarly in IonSource, it was also kind of the result of a merger of a couple of three companies back then and two companies in ZIG that we joined forces very early on and became co-founders of this new company. and it was in the beginning of kind of the mobile app and mobile games ecosystem and we built this end-to-end platform for mobile game developers. The beginning dealt mainly around ad monetization and user acquisition and then it also evolved to game publishing which was to some extent a very big part of the inspiration to what we were building in ZIG.

2:30And it was a very, you know, a journey like that is always, there are a lot of ups and downs. There were days where we thought that it's like the end and there are days where we thought they were going to conquer the world. But I think that we had an amazing team. I mean, there is still, by the way, I want to sit out of Unity today and many are still there. But it was really a unique team and a unique culture. And the values of really making a difference and disrupting an industry was something that really drove us. And I think that we managed to really provide significant value for mobile game developers.

3:18and tell many of them from small to really scaled businesses. And we don't have time, but I have endless stories of game developers who started with game developing being their side job. And then a few years later with us had like, you know, super successful gaming companies worth hundreds of millions of dollars. And there was also the heart of what motivated us with Seek. How do you go to companies in the beginning of their way, identify the potential, and then really put their entire scale on steroids? And when you also put AI and agility capabilities into it and you take it into maybe the most complicated, but also the biggest or one of the biggest terms out there, which is e-com, I think it's a really exciting opportunity.

4:15So I think the journey in I was overall, it was amazing, like the best journey of my life. And obviously taking a company public and being there and, you know, and ringing the bell is a moment that I'll never forget. Right. But I but I think that right now with AI changing everything, it's really it could be more exciting to build new things and things that you can and that you think will really like like this is the era where the new biggest companies in the world are being built. right and you don't it happens like it's not often so i'm i'm really fortunate and appreciative for the journey in that we did in our source and i am super excited about the journey ahead with zig yeah that's that's super interesting i think there's a lot actually that you know i find

5:19Loukas:gaming is is a step ahead when it when it comes to kind of paid acquisition and marketing i think we really see this on uh through through what we do in in um you know paid media buying creative strategy and you know especially when it comes to creative the volume of of ads and you know with app loving and everything that that has been pushed there that that volume model is is translated at a later point to to that of d2c but what talk me through maybe you know when you're when you're talking specifically about a uh a tool that i guess gives you that that insight to to know okay what is working and what's not?

5:56Loukas:How does that work in gaming? What were those signals that gaming companies were looking at? Cool. So let's start with how we did the evaluation in gaming, and then let's think about consumer and D2C. So in gaming, we had game developers approaching us, even in early phases, even when their game was just at the beginning. And then we would run a marketability campaign where we would take the assets that they have, we would add some more assets, whatever is needed for the brand in order to do effective campaigns. We would bring enough users, let's say for meta as an example, and we would measure specific KPIs, right?

6:47So in games, it will be day one retention and cost per install, right? But we needed to bring the audience in a way that it would be statistical. So you need to work on a segmentation and like diversity, but to bring like the right kind of group of customers. And then we would take that number and you would benchmark that compared to a very detailed taxonomy of categories and subcategories that we will maintain. So we will see how these KPIs compare to the KPIs of the subcategory. And then you would run a model of how would it change in scale because obviously these KPIs are changing when you're scaling, sometimes for the better, sometimes for the worse.

7:37But we would model that and eventually we would give it a score and say, do we think that this game can be scaled significantly? And if we thought the answer was yes, then we would literally take everything around scaling end to end. So we would do the advertising and monetization and LTV and analytics, and the game developer would develop mainly in developing the game itself. And SuperSonic, or the publishing business within IronSource, was a very big success. And he became one of the biggest game publishers in the world. And I think that in e-com, there is a very, very strong need for something similar.

8:23Because if I am an entrepreneur with a great consumer product, right, and I just spent two or three years really, really developing great physical product that really makes value, and now I want to turn it into a scaled business, I can go to Amazon, right, and I'll get distribution, but it's very hard to kind of rise above the noise there. I can go to Shopify and I will get the pipes and the store and the ecosystem, but everything is, it's a do it yourself mode. And I can go to, I don't have basically anybody until now that I can go to and say, I'm going to focus on the product, but I need somebody to take care of everything needed around growth.

9:14and we can deep dive around what everything needed is. Part of it is obviously advertising, but it's only one part because there is also building the brand, managing the CRO, the conversion rate optimization, the retention, the inventory fulfillment, the customer support. So we are building this agentic infrastructure that knows how to do most of these things agentically. Obviously, in some areas, we still have humans and really talented humans involved. But eventually, these agentic infrastructures allows us to go to brands. And if we think that the brand can scale, then we can do the majority of the heavy lifting and literally drive scale end to end.

10:04And specifically when we're doing the validation, and I'll say it in high level, and if you want, we can also deep dive into that. But in high level, what we're measuring is what is the CPA, right? So what is the CAC, customer acquisition cost of that product that we are testing? then we are doing an estimation what is going to be the LTV because customers can choose between OTP or subscription and then we're calculating what would be the LTV. And then we are benchmarking that compared to the taxonomy that we now have in e-commerce consumer. And then we give it a score. The score is actually it's much deeper because we go really, really deep.

10:57We actually see, like, we analyze the audience. Was it a wide audience or a narrow audience? We analyze the creatives. Did only a certain creative convert or did different messages convert? Like, we really go deep. And eventually, we have what we're calling the Zieg score, which is our model to predict what are the chances of this brand to succeed. There are also phases before we do that that if you want, we can go. But basically, we have a multi-step funnel that takes about – the test itself goes about a week. Some of it is completely agentic, the beginning of the funnel. And at the end, we have what we're calling the ZIG score that gives us a pretty good validation.

11:45By the way, we've done until today more than 100 of these processes since we started. We started about a year and a half ago. We already have some of these brands scaling and we are already seeing really good correlation between our ability to predict and what we're seeing when we're scaling. So we're starting to feel really comfortable with our ability to predict and scale at this phase.

12:14Loukas:Yeah, it's a super interesting model. I think our job as founders a lot of the time is to mitigate risk. And I think one of the biggest pain points I see from working with so many brands is, you know, they can be, they could have validated a hero product that is doing well. And then they're looking to diversify their SKUs. And through that diversification, you know, especially with Indeed, see a very cash flow constraint. So if you're having to put in a PO at a certain MOQ and that's the wrong direction and decision, then that can be detrimental to a business. So I think this is super interesting from that perspective as well.

12:57Yeah, I think that when you are clearly wrong or clearly right on your product bets, that's actually when it's relatively easier. but the middle ground is the biggest problem because what happens in many times, you're launching a new, whether it's your only hero or a new hero product and you think that your CPA or CAC would be 100 and that's what you need in order to have a healthy, let's say, LTV2 CAC and then you are starting at 150 and now you are in a journey of optimization that you don't know where you're going to get to. You're usually running out of cash flow and you don't have good enough KPIs to raise more money.

13:45So it's a tilt mode that many of the consumer companies are eventually getting into and many of them also don't survive that. But if I'm completely wrong, if I thought something is going to be a cuck of 100 and it's 500, then I know I have no reason to continue to optimize. I can kill it. So I think it's usually, in many cases, the middle ground is actually the most complicated kind of phase for these companies as a general

14:12Loukas:note yeah makes sense and when you're validating these products like how are you how are you isolating the variable to know that it is a hundred percent a product issue and not say a creative issue because we say what you know we'll work with a lot of brands and you can reduce their cack in by half just by producing better creative so what how are you validating and that variable Yeah. So first of all, I'm sure that we're also missing some opportunities. Like, I don't think that there's a system that can verify, and it's not bulletproof, but what we're doing is prior to doing the marketability test, there is actually a phase where we're doing a very, very deep market research.

15:05agentically, of course, and we identify what are the possible angles that we want to go for or that are unique for this specific product. We look at competitors. We look at, you know, we go through all of the different areas that you want to make sense and we really isolate. And then the test itself is checking a few angles. Each of them is approaching a different persona, right? and then it gives us, we are covering several angles, several personas, and we're not looking for, we never think that in those days we're going to get already kind of the full optimized potential. We want to get a score that we know that can get there, right?

15:56So there are different types of extrapolations. But I'm sure that even if you're testing several angles And there might be cases where we will miss a good opportunity. I think it's always, but it's still, from what we're seeing, it improves 10 times your ability to predict compared to just going for it.

16:18Loukas:100%. I mean, yeah, it's great to hear that because, again, one of the biggest things I see with brands and why they work with us is because, you know, They don't know how to diversify their ad creative through things like really kind of going deeper into each of those personas. And with Meta's Andromeda, that's persona driven strategy is even more important. You know, that congruency with pre-click and post-click. So the fact that you are spreading that to get as accurate data as possible makes a ton of sense. So you've worked, you know, 100, 150 brands that you've run this model on through Zig.

16:54Loukas:Like what are some of the biggest trends that you're seeing with the brands that, you know, have validated products that really work and those that don't? Yes. So I think the most interesting kind of trend, and it's also something that I think we can learn from for mobile games as well, is that we all know that media, meta, but not only meta, is becoming really expensive. I think the biggest pain that you need to deal with every day, I assume, with your customers and everybody in our ecosystem is that media is becoming extremely expensive. And I think that what that creates is that most D2C companies, unless their AOV, their average order value is like$500 ,000, most of them are actually losing on the first purchase, and they need to optimize towards LTV.

17:53And that is something which is everybody in games are doing it, right? Mobile games, by definition, are losing on the first customer. Even sometimes they'll be 0.1 or 0.2 in kind of the first few days, and they're optimizing on a lifetime value, and they're building models to predict lifetime value. We are doing the same in e-commerce, right? And I think that we're going to see more and more consumer companies, especially ones that will be sophisticated with data, that are going to predict lifetime value. So understand how many repeat purchases they're going to see, or if it's a subscription, how many months they're going to stay subscribed.

18:42And to be able to predict that early on. By the way, with AI, it's easier to build really strong predictions here. And then you want to also optimize towards those predictions. And if you know how to do it in an accurate enough way, then you have a very big advantage. Because then you can bid higher and you can be more aggressive and you can scale much faster. And in the heart of what we're doing and the way we're scaling is that capability. We know today very early on to predict LTV. We have a very strong trust in our algorithms, and then we optimize towards that LTV. By the way, the challenge here is that short term, it creates a cash flow issue, right?

19:32Because if you want to scale, so we offer that financing as well as part of our solution to our partners, right? I think that's a really, by the way, that to me, you know, I mean, that proprietary model in terms of being able to score and predict the success of a product is one thing.

19:52Loukas:But I think being able to see everything in a centralized location in a world where there's so much saturation in terms of tools that you can use, we are, you know, we're spoiled with choice. and as a result, people fall into complexity bias. I think it really helps take a step back and just centralize things. So I think that's a really interesting part about the product. Thank you. So what I would love to just go a bit deeper, because it's 100 % true, like CAC to LTV, you see the brands like Grunz that have just sold for over a billion and we're starting to see these really big DTC acquisitions happening again.

20:34Loukas:And I think, you know, a big part of that is subscription, right? I think the valuation of a business can transform with that model. I think the other thing that I personally have seen when I started my brand about a year ago is the importance of gross profit. You know, even when you are developing a skew, if the unit economics aren't strong enough and you're even below 70, 75 percent, you're going to really struggle. So absolutely, absolutely. When we are checking the potential, we have two kind of minimum thresholds. We think that they are the basic in order to have a credit you can scale. One of them is exactly what you said.

21:17You need a 70 % minimum gross margin. But when I say margin, it's landed cogs. Like it's including the delivery and pick and pack and everything. So you need to have, let's say, 65 % to 70 % compared to Lend and Cogs margin is fundamental. And you need to have, I would say, an entry point of 2.5 kind of LTV to CAC prediction ratio. If you have these two, then you can scale. so these are two really really important threshold when you are evaluating

21:58Loukas:the potential of a brand yeah super interesting and I think you've spoken a little bit about AG1 previously and I would love to because you know there's there's one we see it a lot where a lot of the brands they kind of scale incredibly quickly they'll they'll get to spending you know multiple million pounds a month through Meta. And they start, you know, having to shift focus to really become a true brand. Some of them do it much earlier, but others are more scrappy and direct response driven until that point. What do you think creates true defensibility for an e-commerce brand? Like what creates that moat?

22:37Loukas:Yeah, I think that first of all, you really need to have a good product. Like I think that I'll tell you what's really hard, and I almost feel that it's unfair to expect it from entrepreneurs. I think that if you want to really scale a brand D to C, you need to be A plus in everything. You need to have a really good product that provides real value, right? That needs to be deep. You need to build the best performance marketing machine. You need to be able to get financing because you can scale without having a pretty big dip in the road. And it's almost impossible because most of the people, they're either really, really good product people, they're either marketing.

23:35It's very hard to raise money for consumer companies. So I think it's really hard to do everything. And that's why we're trying to find those who are the best at creating the products and we can really help them with everything else. I think that, I think, and just also to refer to what you said before, when you're looking at Groons and you're looking at AG1, and I don't know if you saw, but IMA, the public, so you can also see all of their details, super interesting. I think that when these companies, by the way, I don't know if you saw, I think, if I remember correctly, I think IM8, they took about a billion dollars of CAC financing, something like that.

24:20Yeah, it's crazy.

24:20Loukas:We had Danny on the podcast. Yeah. And I think that what we are trying to do, those companies, when they have enough cohorts data, they allow themselves to be really aggressive. aggressive, right? So, so they can also, they can also live peacefully with a$300 cock and still scale, even though they started with 150. And, and if you can do it when you're starting to scale, right? And you can trust kind of the PLTV algorithms, then you can really increase your chances. Right? I think that, I think that, and then, so if you can do everything really, really well, you have a pretty strong moat because it's very hard to also have a great product and the marketing machine and everything.

25:16So I say that if you're A plus in everything, it's a pretty defensible moat. I think in AG1 and also IM8 and others, obviously they're very like also celebrity heavy. So that's another angle, right? when you're adding people that are adding trust and reduce tax significantly. So that's a strategy of its own. I think Grooms, by the way, is an extremely interesting case study because Grooms, what they did is they obviously were inspired by AG1 and they thought, how do we give the same value but in a much easier kind of delivery method to consume, right? So instead of consuming this daily powder, which is not really fun, I would turn it into this daily sachet of six gummies.

26:14They would give me, by the way, it's six gummies because you need to have enough of the active dose. And gummies, they only have a small place for dose. Very hard, yeah. And it's crazy to think that that innovation around delivery method created that insane growth. And I'm sure that there are so many companies out there that are thinking of some innovation, whether it's in the delivery method, in the formula, in the way it looks, that they don't even know that their specific innovation is really making a big difference. and now they can use it to scale significantly. And obviously, this is one of the things that we're trying to find out really soon.

27:00Loukas:I completely agree. It's something we're seeing a lot with the brands that are really exploding. They've almost allowed other businesses that use a different format to deliver that same outcome to, I guess, like invest in the education around almost that whole industry or that whole sub-niche. and then they come in with a new format new means of delivering the the product um and that's that's you know massively working for a lot of brands so yeah i i completely agree with that what would you so you know let's let's say we've got we've got bigger brand owners that listen to this podcast a lot of new brand owners as well and i want to make sure we're covering some some value there because i know you will have it but with maybe the smaller brand owners that aren't quite at the level where they would use something like Zig.

27:51Loukas:What advice do you have if they're at that point in their business where they're kind of breaking through seven-figure annual revenue and maybe they're at that point where, you know, they're thinking about diversifying, bringing out another SKU. What should they be thinking about to set themselves up for success and how should they approach launching that next product? Yeah. So first of all, just to clarify, We do have also bigger customers, but we actually are mainly working with customers who are generating like 2, 5, 10 million a year. And we think we can scale them 10x in two or three years.

28:33And we're seeing some great kind of beginning of case studies around that. But we also work with bigger companies. I think that when you want to expand your portfolio, I think you should use the exact same method of doing evaluation first before you really commit on inventory and full launch. And by the way, it's even much easier to do when you already have a brand and you're already selling at seven or eight figures. If you now want to test, doing this type of test where you literally generate all of the brand assets that you need in order to evaluate what's the interest. And you can drive customers to that page and you can measure and you can even compare that to the convergence you're seeing on other products.

29:27I think that it's a very easy and very effective way to decide what products you want to bet on. Obviously, one of the biggest challenges in e-com is inventory, and especially around inventory planning. And sometimes the leeway can be two months, sometimes it can be six months. so it's it's very very if you can repredict what is going to be kind of the the demand here then it's a life changer um i do think by the way um that long term you don't want to you don't want to be dependent on on a single hero so i do think that you constantly need to think about how can you launch new product by the way a product can be an add-on a bundle or a product can be of course an additional hero.