Signals cut down, marketing stacks bloated. How to trim waste and skim across privacy hurdles?
Why you should listen
Everyone’s reacting to privacy. Aman Sareen says we need to build proactively for it.
In this episode, the CEO of Aarki shares what it takes to grow in a post-IDFA, signal-starved world and why most marketers are still playing defence. From SKAN constraints and cloud cost squeezes to the blind spots between user acquisition and retention, Aman breaks down the real obstacles to efficient growth, and how to build a stack that’s ready for whatever Apple, Google, or the next regulation throws your way.
It is all about calling out the inefficiencies that performance marketers silently put up with, and replacing them with strategy that scales.
Top 3 takeaways:
- Rapidly changing privacy norms aren’t your enemies. Waste is.
Apple’s privacy framework isn’t going anywhere, so stop hacking around it and start building for it. Aman shows how understanding signal delay and attribution logic can sharpen your ROAS, not stall it. - User acquisition without retention is a leaky funnel.
You don’t need a new vendor; you need a flywheel. Retargeting data can supercharge UA models. Run them in silos and you’re burning budget. - Own your stack or keep paying the toll.
Renting tech = rented outcomes. Aman breaks down why owning infra (from data centers to decisioning engines) is the fastest path to performance, control, and long-term margin.
The transcript
Peggy: So here we are at MAU, and it’s a year later, and I’m speaking with you again. Aman, Aman Sareen, of course, CEO at Aarki. And we were talking about what’s next after privacy. And here we are, because it’s still the topic, it’s still the buzz. And long before it was cool you were addressing it. So, it’s, first of all, it’s just great to have you because you love solving problems, you love deep diving into this, and you have, yeah, you have built something that addresses it head-on. It’s great to have you.
Aman: Thank you for having me, Peggy. And exciting to be talking to you again. Exactly. A year before, uh, I think today we relaunched Aarki with you. So, thank you for having me again. I’m excited to talk about all the privacy challenges that marketers face today.
Peggy: Well, it’s understanding those privacy challenges that has to come into what you build, what you enable. So, let’s start with that. You know, the match between the challenges and what you have built in Encore. What did you see as what you needed to address and how are you addressing those with Encore?
Aman: So, look, if you think about privacy, and let’s focus on Apple for a second, okay? Um, this is not a conversation of today. This conversation has been going on for almost five years now when Apple, uh, launched the ATT framework.
Peggy: Mm.
Aman: Um, and I think because of that, lots of businesses, big and small, were impacted, including large social networks whose stock prices kind of tanked because they couldn’t figure out what was going on. So, this conversation is, has been building up since then. And if, if you look at the, if you look at the world today, I think we can divide the world into two parts, right?
One is all the solutions that our friends at MMPs built to solve the privacy challenges. And I think those are great solutions for the categories that work with them. And then there’s the other part of the world, which is, I would say more regulated from a privacy framework standpoint. Right. Uh, categories, uh, like social networks, uh, healthcare, fintech where payments are involved.
I think those are very privacy centric and at the same time, they may be also much more integrated with the Apple ecosystem. So they want to do what Apple thinks is the best. Right. So, there’s two worlds right now. As we, as we look at the world today, what we wanted to do at Aarki was to build a solution which addresses both of these problems.
And that’s the premise behind why we built Encore.
Peggy: Well, you started it, you teed it up, addressing those problems, addressing them head on. Explain the purpose of what it’s addressing. What is it addressing specifically?
Aman: There’s a couple of things to look at. One is, uh, you are solving problems for signal loss. When I say signal loss, what I mean is when, when you have ATT where you cannot track users, marketers still have to find those users, right? Especially in the Apple ecosystem, where all the wheels live, if you will, for if you use the gaming world terminology.
Right? So how do you, how do you then make sure that when you spend money for a client to get those users, it is actually well spent, not wasted, and. there’s a positive ROAS for the clients, so, right. So that’s one challenge.
Now, the second world, and I think if you look at the categories, right, so the gaming categories solve it very well.
They rely on the MMPs, and that’s the ground truth for them. And how they look at this world is very, very simple. If Platform A works very well because the MMP says so they will put more money on it. If Platform B says it works very well, they will put money, more money on it. And I think there’s nothing wrong with that.
But then there’s the other side of the uh, solution, which is actually more pure, which is what Apple gives. It is SKAN. What SKAN does it, it makes the playing field level for everybody, right? There is no, I cannot adulterate it by using my black box MMP algorithm. There is no magic attribution that I don’t know as a platform or you don’t know as a, you know, another platform.
Peggy: Mm-hmm.
Aman: Um. So that’s the solution, right? How do you build solutions where there’s very low signal? There’s attribution delay, right? Because Apple kind of delays the attribution so that there’s more privacy friendly for the clients, right? And then how do you build solutions which bring the same type of ROAS that the MMP solutions bring?
So, I think that’s what we are solving for here.
Peggy: You’re building for that today, but interestingly, everything is shifting. So, what you build at the moment, you know, might not be what’s next, but you’re bridging that. How did you shape that? Through understanding your, your customers and their challenges. But you know, you had to choose, am I building for the challenges that they state, or am I looking at what’s coming next?
Aman: So, I think, uh. Encore is not, you know, something we dreamed up like yesterday. You know, funny thing is, I think not a lot of people know this, but I built one of the first performance DSPs in 2012, like 15 years ago.
Peggy: Before it was cool.
Aman: Before people knew what performance yes is. They were no MMPs, okay. Um, so I think things have come full circle, if you will.
Right. And, uh, if you think about what the future looks like, I think what we want to do at Aarki is we wanna be very privacy centric. We don’t know what will happen tomorrow with. Let’s say Android. Is Android going to actually, uh, start using their privacy sandbox a hundred percent? We don’t know the answer to that.
They may, they may not. Um, so we have built a very agnostic ecosystem where we can use any privacy framework. It fits into our, um, you know, Apple privacy framework. Google privacy framework, they all fit into our system in a very normalized fashion, and we can keep building solutions on them to target both sides of the equation.
I don’t think building a very specific solution today makes a lot of sense. If it’s not future proof and all the investments we are doing is making sure that everything we do is future proof; Android could build any system, we will be ready for it. Apple can make changes. We will be ready for it.
Peggy: I mean, was the idea that, that we wouldn’t know really how privacy was going to swing, but we knew that it was going to be a constant.
Aman: Yes. Privacy is, privacy is gonna become more and more user friendly.
I do believe users will have more say into what data they wanna share and not share in the future. To be honest with you, I don’t know if Apple does a very good job at letting the user select what they want to do. I would actually like to get more, better offers as a, as a user, right? Like I don’t want my ads to be random. Now with Apple privacy framework. The users actually have a very difficult time opting in. Opt out is normal, right? Yeah. But my hope is that in the future, the privacy frameworks are more user friendly from a choice standpoint, right? I don’t think you get choices today. You get zero or one.
Peggy: Yes.
Aman: And for tech-savvy people like you and me, we can go and say, you know what, no, no. Show me better ads so I can actually find, find brands we hopefully I will love in the future. Or know about products that my beloved brands are selling to me. Right. I think that change will happen and I think with that, I think there’ll be a good balance of privacy and user choices.
Peggy: That’s a very interesting point because that is that value proposition that I’m hearing more and more, that you know, it’s an understanding that I want brands to step up in a certain way, or I want to hear from them in a certain way, or there is some value to what they have and that relevancy, that is the value and that is what users don’t have. That is what consumers don’t have.
Aman: A hundred percent. Not only that, if you think of the, the wall gardens, I think they do a very good job at product discovery for users, because they have all the information, they don’t need to rely on Apple, uh, or, or any other platform.
Peggy: Yeah.
Aman: But for the open internet, which is like more than 50% or maybe more of the volume. Yes. There, there is no choice, right? Mm-hmm. You are stuck with whatever, uh, a large tech company tells you, right? So I think that that will change eventually.
Peggy: You built this. Into Encore, it’s baked into Encore with this in mind that there’s going to be an interest in choice, that there’s going to be an interest in value, and that privacy is a constant. What about the marketers?
I mean, you’ve had your first demos, you’ve had your first reactions to it. Are they seeing it that way, or are they looking at it more strategically and maybe not seeing the bigger privacy picture?
Aman: So I think when we look at Encore as a solution for marketers, it’s a full funnel solution.
Peggy: Mm-hmm.
Aman: And very privacy centric.
Right. So I think I can say that it’s one of the first or one of the few privacy friendly, privacy centric, full funnel platform, which not only helps you acquire new users, but also helps you retain existing users. Yes. I think retention is something which is gonna become more important and top of mind as we move forward.
I think we have talked about at you and I, yes, at length.
Peggy: From Day one. Mm-hmm.
Aman: Keeping your user in your ecosystem is very important. And I think there are only so many users who spend money, right. For gaming, for example, right. Now and there’s so much competition for the, that mind share, not only from mobile games, but also from, you know, your connected TV, right?
There’s, there’s a war of mind share. So, I think, so we built Encore so that you can do all of it together at the same time. So, I think what, what, what clients are surprised they talk to us is that “Oh wow. You can do everything in one platform. So, I don’t have to go to three different vendors to accomplish my marketing goals.”
Right. And then for the others it’s more of an education. Other thing, what we, you know, what we do is we have our own data centers. Okay. We are not on cloud. We own and operate four data centers across the world: Two in, you know, America, on East and West coast.
We have a data center in Europe so that we are fully GDPR compliant in Netherlands actually. And then we have a data center in, in Hong Kong for all of APAC traffic. Um, and I think if you combine what we are doing, you know, with infrastructure and investment and infrastructure, um, I think we can move much faster than most companies on cloud.
Our software development is much faster. We, we control the machines. It’s a very sophisticated hardware-software solution. Um, and on, on top of it, we can listen to all the ad calls of the world. Right. We are listening to 5, 6 million QPS, which is literally the fire hose of the world. And based, because we have so much data, we can actually build very interesting solutions for different use cases, for different verticals for our clients. So, I think if I, if I encapsulate it to a client, they’re like, “Oh wow. Like we didn’t know that you have all these solutions and uh, and we can actually just work with you to accomplish a lot of our marketing goals.”
Peggy: Not only is it more effective, but it’s also when I’m talking to marketers, they’re saying, well, that’s also very cost effective because they’re worried about that; they’re worried about being squeezed by the margins
Aman: So we don’t worry about margins too much because of cloud costs going up because our scale is going up. Yeah. ’cause you don’t have to worry about that. I don’t have to worry about it. Yes. Um, uh, and you know, one of my pet peeves, and I think again, you and I have had these conversations, we want to remove all the middlemen possible from this ecosystem because as much as dollar should go from the advertiser to the publisher. Right. And I do think that cloud is squeezing those dollars out.
Peggy: Yeah. Yeah. That’s exactly what I’ve heard. Yeah. But the inspiration for layering UA and retargeting. I mean, that’s a vision of full-funnel marketing that many marketers don’t yet really, truly have. They’re saying we wanna grow, but they’re not really defining that.
What was the inspiration for, for you at Aarki? Because of course you have your street cred in retargeting, but then layering on UA, that means you have something else in mind.
Aman: If you have a unified data lake where you can see the full user journey from acquisition into actions into then maybe dormancy, hopefully not, right? Yeah. But then reactivation.
So, I think that funnel works really well if you’re doing everything at the same time. Uh, like we have many, many clients who work with us on a very consultative basis because they’re like, okay. Okay, you help me get these users, or I have this huge pool of users, I don’t know what to do with them, right?
So we always tell them, “Hey, if you have a huge enough app, if you have never tried remarketing, let’s get those users back. It’s cheaper.”
Peggy: Mm-hmm.
Aman: Like it’s a tenth of the cost to reactivate a user versus get a new user. And you know, you have users who have purchased, right? You have all the history. Let’s work together and make sure that we get those users back.
Peggy: And those models, they also are important in UA. Again, because you’re saying these are the types of users that you have, you’re getting them back. Let’s turn that around and make that part of the flywheel also for UA.
Aman: A hundred percent. So if you are working with us on remarketing, our UA models are already learning, right? Yeah. And then that, again, it goes back to wasted media spend. When you have all this learning shared across the two strategies, there’s much less wasted, media spend.
Peggy: You have so many different customers that you’re working with and you talk about the effectiveness of this in powering that flywheel. Do you have an example to share?
Aman: So we have, uh, a retail client. Um, and, uh, they, you know, what they do is they have like drop sales, right? They, they do drop sales over the weekend, and what they do is they SMS those users saying that, “Hey, we have a sale, and come and check it out.” Right. And they spend, let’s say like 50 or a hundred thousand dollars per month on that SMS campaign. Uh, when we started working with them and they told us this, and we asked them like, how do you track it? You know, like, what’s the tracking mechanism?
And it was very low tracking ability for those SMS campaigns. It’s like offline, online. It’s complicated. Yeah. Um, so we said, why don’t we do this? Let us, once you SMS these users, let us help them push a message out on, okay, in real time, right as they’re browsing or doing or playing a game or browsing the internet, let them, let us send them a message.
So, we saw a 3X higher ROI than the SMS campaign on those remarketing messages. Eventually we started working with them where they retired the SMS message campaign and now, they’re exclusively working with us and we can actually pinpoint, target, and show them the right, like we can show the right product to the right audience,
Peggy: okay?
Aman: In real time. Over the weekend. We call it a blast campaign, but it’s not blast. It’s actually, uh, you know, a precision blast campaign. Okay. So that’s a real example of how we work with, uh, somebody to achieve like massive gains on their investment.
Peggy: And of course that is also feeding back into the AI, into the models.
Aman: Yes, and it feeds back into the user acquisition model. It gets them the right type of users, and then we see those users, what purchases they have made, what products are interested in, and then we push them back into the retail funnel, if you will.
Peggy: So I just had the opportunity to talk to Wangxin [Li], who is of course, you know, a data science architect at Aarki building this. And what we were excited about and discussing, because we were actually referring back to what you had said, it’s about orchestration, it’s about taking these decisions that are for marketers, be it bids, be it creative, and, and orchestrating, yet letting the AI sort of bring all of those capabilities together, and that’s the orchestration that is enabled through Encore.
But the next frontier is when that’s more autonomous, when it’s not just taking out the heavy lifting of making the decisions, but really making the calls so that the marketer. Yeah, that there are more automated workflows. What does that look like?
What is, you know, when do we get there? Is that, is that the end game for Aarki?
Aman: Sure. So, I think, I think the end game for Aarki is very simple. I want to equip all the marketers who work with us with an Iron Man suit.
Peggy: Mm-hmm.
Aman: Okay. So that they can be 10X, 100X more productive. Okay. The goal is not to remove humans from the equation, but the goal is to make them extremely productive so they can do things which the computers are unable to do.
You think create, you know, ideate, right? So as we were building Encore, our vision was that Encore will be an agentic product. It’ll be an agent which will become part of the marketer’s team. And it’ll help them execute complicated tasks with human supervision. Okay?
AI or machine learning in marketing. Some people think it’s just user evaluation, figuring out how good or bad the user will be, but it’s much more than that. It’s finding out the value of the user to an advertiser, making sure they get the right offer. Finding that user in the right context and also paying the right price for the ad. Right. It goes back to this margin equation, right? Yes. So, you have to do all these things at scale, 4 million times a second.
Okay. What we have done is we have incorporated lots of these agentic products in like, let’s talk about creative, okay? We have, uh, creative testing is complicated these days. It takes two weeks, four weeks to create a new concept storyboard, all that stuff, right? So what Encore does is it spits out lots of features of winning creatives right into a GPT model, which creates new storyboard concepts, and our creative team can rapidly test them out, right?
So, goal is maybe by sometime next year we will start offering this as internal agent to our internal analytics team. And then once we are happy with it, we’ll start doing some lightweight data testing, but it’ll happen sometime next year.
Peggy: That’ll be exciting.
Aman: Yeah.
Peggy: It’ll be exciting because that will push the boundaries. I mean, it’s not just about which creative is effective, but you want the AI to select the right users in the first place to show them to. And so it’s not just, you know, our discussion right now is still around, oh, it’ll help create the winning creative, you know, we’ll, we’ll get, you know, a thousand times faster. It’ll spit out all of these different versions. It’s more about speed than selection. And selection is where you’re gonna see the efficiencies again.
Aman: Exactly. So I think, yes, you can create a hundred creatives, a thousand creatives. Some platforms may tell you they can do that. Yeah. But then what? Right. But then what? It’s the orchestration of the entire thing, getting the one or two out of the thousand and then repurposing the whole engine. Like again, it’s a flywheel.
Peggy: Yeah.
Aman: Yeah.
Peggy: And there are also some balances here because you can say, okay, I’m going to get, for example, I don’t know, something very, very low cost.
But then if you’re looking at the full funnel, cheap can be expensive.
Aman: Yes. Yes, you’re right. I don’t think it’s about cost, it’s about efficiency. It’s about quality and it’s about, again, you know, I say that we live and by die, by ROAS in this industry and it’s all about that.
Peggy: Well, I do wanna have some geeky questions for those who want them and rapid-fire questions. Minimum integration time. We’re talking about all these benefits from Encore. How long is it gonna take?
Aman: Sure. So, I think if somebody does not work with us and they wanna start working with us, I think it’s probably two to three weeks of integration time.
Peggy: Okay.
Aman: It’s not because they cannot integrate fast, it’s because we want to get the fire hose of data from them before we start working. So that fire hose takes two to three weeks to get, you know, the right data and signals to our model so that we are more efficient in our media spend.
Peggy: The new privacy regulations, they’re coming hard and fast. That’s not going to stop. How do you handle them? How do you adapt? I mean, you built it from the ground up to be privacy-first when people weren’t even certain what privacy was going to look like and you know, the flavor of the day was changing every time.
Aman: So we have created our own Aarki version of privacy framework internally. And the things that SKAN does, I’m assuming like, you know, like things that Google sandbox, privacy Sandbox does, we plug in all those things into our, our little framework and we can be future proof because of that. So, let’s say if somebody else comes up with another privacy frameworks, I doubt we haven’t thought about everything, but if we haven’t, we can enhance that schema and add to it. So that’s how we are taking care of it.
Peggy: Mm-hmm. So there’s a lot of opportunity, a lot of choice. And marketers are looking at the AI platforms, the AI offers, even just the single point solution tools and asking themselves questions. What is the one question that they should be asking or maybe aren’t asking when they’re vetting these platforms?
Aman: Yeah, I think I, I say that to everybody. Make sure you work with a platform, not a reseller of other platforms.
Peggy: Ha Okay.
Aman: Okay. Very important, right? Rented technology versus own and operate technology. I think that’s the key to ask for.
Peggy: And that’s what you’ve done from Day 1 because you have your data centers, you have your tool. And you have the methodology to teach it and to keep it up to date. So yes.
Aman, always a pleasure to speak with you because you have seen this, you built it, and uh, you see what’s coming next. And one of the exciting things will be autonomous systems and agentic systems coming very, very soon.
So, I’ll be watching for that. Thanks again for taking the time.
Aman: Thank you for having me, Peggy. And hopefully next time we can talk about all these agentic systems.
Peggy: Yeah, I love agentic systems. They are super cool. Awesome. And that is what takes all the heavy lifting out of these difficult decisions.
And yes, it’ll be a pleasure. I look forward to it.
Aman: Thank you.
EN 
KO 
ZH