Automate User Acquisition with AI

Oz Silahtar, CEO and co-founder of UAhero was on The Hive to talk about automating user acquisition with AI.

Oz shares some of his superpowers with us all and also gives his predictions about what the UA landscape will look like in 12 months.

Transcript

Peggy: What makes a hero?

Well immediately you think of maybe capes, armour, superheroes... Well you're not off the mark entirely, but there are other types of heroes. The heroes we're looking at today are the makers, the doers, the shakers, the thinkers, the innovators... In other words, the super brains who can supercharge your UA efforts.

This is The Hive and although my guest wouldn't call himself a superhero or maybe even a super brain, he has certainly got the cred. He has a degree in electrical engineering and computer science from MIT, a long track record in R&D and biz dev at companies, including Qualcomm, Intel, you know the big names. And today he is the co-founder and CEO at UA Hero where he and his other super brains are working continually on ways to rocket your UA and we're going to find out how, because this is The Hive and my guest today is Oz Silahtar, as I said, CEO co-founder at UA Hero.

Great to have you, Oz.

Oz: Hi, nice to see you, Peggy.

Peggy: Why don't we start by hearing that origin story? Tell us about UA Hero and what the company is.

Oz: Basically what we do is all the user acquisition campaign management with AI and as a result we wanted the name to be easy to understand and stands for user acquisition and hero because, as you said, our platform gives the user acquisition managers super powers, lets them do everything better, faster, and consistently.

We do lots of predictions, forecasting, kind of like seeing the future we could say based on past data.

That's one aspect and the other aspect is consistently doing the right decisions. Kind of like a super brain, as you said, when you manage your campaigns for user acquisition.

Peggy: When you start a company you don't just do it for fun. You might have because you have a degree at MIT, so you've got to be a tinkerer at some level, you know? How do I do this? How do I make it better? But you also wanted to address a pain point. There's a need in the industry. What need are you addressing?

Oz: This is a very good problem for machine learning AI because user acquisition managers need to manage lots of different campaigns on different channels, different geographies, different source apps. It has been proven that if you do it data driven, then you get better results. Your campaigns are more optimal, and you get more users, more profitable users. So we saw that this is challenging. This is challenging because good UA managers are not that many. I think the number of UA managers is less than the number of gaming companies in the world, probably. And the second is to do it consistently without getting tired. It's a perfect job for AI.

Peggy: We talked about the what of your company, the why. Why you did it. Now we talk about the how. Is this about making more intelligent media buys? More compelling creatives? All of the above?

Oz: Kind of all of the above. There's two aspects. One is the bid budget management of the campaigns and the second part is the creative management which includes getting the most insights from the data on how to manage the creatives. Which creatives to replace? What to work on and a lot of automation on uploading the creatives and such. Our tool is to help the UA manager so they can be more creative and strategize, not to replace them. But the work that typically needs to be done manually or even with the help of Excel or some kind of calculations they should let the machine do it and they can strategize on how to be more creative, what to do. And we empower the UA manager.

Peggy: Which is right aligned with the UA Hero. You allow them to be the heroes here.

Tell me a little bit about how you work with companies. Maybe some examples. Who have you worked with? Who are you working with?

Oz: We work with some hyper-casual publishers, casual gaming companies, mostly. Basically a gaming company that would publish their own game or the publisher.

The onboarding is very easy because we use APIs of different networks. Right now we are integrated to Facebook, Tik Tok, Snapchat, iOS, Google, Unity, AppLovin and we also connect to the M&Ps using the APIs, no SDKs necessary.

So the start of the work is... onboarding is very easy just like like one phone session and getting the right information.

At this stage we're a startup. We have very close relationship with the customers and I would say at least once a week, multiple times a week, we have calls to make sure that everything's going smoothly and get feedback.

Peggy: So how long would it take or what is the process to sort of, you know, teach the AI? Everything is about teaching the AI. First it has to figure out how I do things, then it can offer really valuable sort of recommendations. What does that process look like?

Oz: Well, it's all automated so basically once we connect to the data of the gaming company, then that evening in offline the AI or the machine learning algorithms do their training and what we would call the learning. Invisible to the...

Peggy: … Invisible and immediate.

Oz: Yeah! Yes.

Peggy: Sounding like all those superpowers again, Oz. I love it.

Let's talk about what marketers are able to do if they can up their game with the help of AI. Because, of course, it's a privacy first world. What do marketers need to know and where does marketing automation help them?

Oz: The usual goal of the gaming marketing person is to get as many users as possible, as long as they are profitable. A tool like ours allows us to exactly do this.

So, basically optimizing every single different, at a granular level, for every campaign and network, different geographies to make sure that all these decisions are made optimally and as many users as possible are acquired for this game.

Peggy: What are you seeing the big guys, the companies that are punching above their weight, the companies that are really making their mark, what are they doing at the cutting edge of UA? What are you seeing?

Oz: I think they are fast at adopting and trying different networks, different channels. I think the first thing they do is, as the systems change, scale what worked, such as ask Google's new changes, they adapt to the new systems like the conversion value, learn how to use it so that they can get the best data.

They're exploring different networks, different UA channels so that as the environment is changing and the performance of different networks is changing, they can maximally utilize them.

Peggy: Okay, so they're experimenting, they're getting on top of new channels, they're staying up to date with the changes. That's what they're doing, that's what we need to aim for.

Now, many companies, smaller teams, they might be low on resources, low on budget ,or again, just a smaller studio. So how can they apply this to their own UA campaigns and even possibly working with UA Hero?

Oz: I think if a small team, of course, is more challenging and that's exactly what we are trying to do, help them, to help them. Once they are using our system, they'll have a lot more time to think about these strategies to follow and if they want to onboard a new ad network channel or to manage it, they don't need to spend time.

Basically our system can work fully automated. Our AI shows all the different configurations, all the different decisions in an easy way and all it takes for the UA manager to approve or okay or just fully automate. So this way they will have more time for staying up to date and we are also staying up to date. So we have regular updates to them, we have regular calls, and our system is always adapting and, you know, getting the best performance.

Peggy: So you're working with smaller studios which need bigger wins, to tell you the truth, because of the size, because of the competition in the marketplace. Maybe you can walk me through an example of what a small studio did to get the biggest uplift with you.

Oz: I mean we had a case where a small studio had a game which you know had its days and kind of was going down in performance. We basically managed to find the channels where and the geographies where the game could still be popular and by adjusting the bids and the budgets accordingly we extend the lifetime of the game and that way they got a higher revenue. Effectively I think they had multiples of the revenue from that game.

Peggy: I hear that a lot in the industry, you know. There is some low hanging fruit that many games companies don't see. It's something like maybe a different geo, many times just localize everything and then you'll get these different opportunities.

Let's stay on opportunities. Which are the game genres that have the most potential for profitable UA and the highest long-term company value, in your opinion?

Oz: With genre we don't really make a distinction, I think. I think once the game is ready for its metrics are, like essential metrics, are working well and it's ready for global launch for mass UA, I think uh it will benefit. So, the trade-off is different between hyper-casual and casuals and all different sub-genres but, I think they can all benefit.

Peggy: To your point, you know there's a lot of blending out there as well. I'm seeing casual is embracing storytelling, hyper casuals trying to look like casual, a lot of movement going on out there. What do you see? How has that changed? How's that impacted what you offer?

Oz: So the key thing here when you do bid budget management is the, I mean of course you have to adjust your bids and budgets based on your LTV, and that's like the essence of the algorithms. And when you do LTV modeling there are different kinds of revenue.

Like the hyper casuals traditionally have advertising revenue, casual games traditionally have in-app purchases, but as the genres blend, the revenue models blend. Our model works with any combination of these and predicts their LTVs.

Peggy: Staying with some best practice some rules of thumb, what can you offer? Because usually in these systems, when I interview companies like yours, you're ingesting a lot of data. What's the data telling you about the state of UA?

Oz: So for each game we use that game's data for the AI to get trained and to learn and because the different UA tags for the game doesn't necessarily translate to everyone anyway, so when we work with a game company all the decisions are based on that game company's, that game's, historical data.

Peggy: So it's very individual, but you're talking about historical data. Just like to zoom out a little bit to the modeling because it gets really exciting when we can predict. Looking at historical data, Covid told us "forget most of your models because those historical models need a revamp" basically. We need to adapt and predict your modeling. What can I do with those models?

Oz: All the key metrics we are predicting and the top ones that come up to mind is LTV, retention, and volume. There are others like the output and others that need to be tracked. We actually show all of our predictions and along with our bid and budget recommendations which can be applied as I said automatically so that the UA managers, they can actually if they want they can go and understand the predictions that actually led to these decisions.

That way if there's something in question we can go investigate and if there's an anomaly of some sort we can make changes because the AI is based on historical data, the data that you should give them basically, give the algorithm, there can be instances where maybe the UA manager knows something that's not possible to know, such as some tip from the network or some other thing that's happening. So in those cases they can go and modify these decisions and make it more adaptable.

Peggy: There's that human element or hunch even that you can bring into the equation. As you said yourself, everything's changing and it changes with you.

I want to stay with those changes and even end on those changes, because we've hit into a time where privacy is challenging, so is the economy. Reading over MobileDevMemo and Eric Seufert, you know, ad tech companies, like Adjust, trimming head count, studios freezing projects, laying off employees.

What's your take on the UA landscape right now and how it might look in say 12 months?

Oz: The need for UA managers and the UA activity will not change. I would say some kind of more retention campaigns are already important. They may become even more important because once you acquire the users you want to maintain, like keep them as long as possible.

I think the world is changing but the needs are the same. You still need to get more users for your game, for your app, and that will not change. And like I said we want to empower the UA manager and let them be creative and strategize instead of doing tasks that the machine can do for them. Let machine learning do what it can and let the UA managers be more creative.

Peggy: You said it yourself, so I have to follow up, Oz. You know we talked about UA but then you're saying "well retention is going to be more important going forward"... makes a lot of sense of course because you want to acquire users, you want to get the most mileage. You can't just keep re-acquiring the same users for that matter. What about making me or a marketer into a retention hero? Where does UA Hero come in there? Is that part of what you can equip me to do or will be equipping me to do better?

Oz: It's possible in the future. Even now as I said we predict retention and LTV. The higher the LTV users higher retention users. So already doing the UA successfully and better helps with that. That's a good start.

Peggy: Okay, Oz. It's a cliffhanger. You don't tell us exactly what's coming, but we know something is coming. It's almost like Netflix. We have to tune in later on that and talking about what's coming,I know that one thing that's really important we didn't touch upon it, creative optimization. That's a big thing that marketing automation does. What can we expect from UA Hero there?

Oz: Well, we are already working on it and we have shared with certain customers our first versions on creative management. Like I said our philosophy is to first come up with bid budget management and now creative management.

It has a bunch of sub-features that I won't go into now, but the initial version we are showing some customers to get feedback to make sure that it will be very usable and nice. Soon to be launched, very soon.

Peggy: Okay, so we've got bid management down, retention coming, creative management already in beta, already shipping, already in front of your customers. A lot of reasons to check in with UA Hero, and I'm sure people want to do that. What's the best way to do that? How can our audience connect and continue the conversation with you, Oz?

Oz: I think we are open with all channels, our websites uahero.ai. Through our partners, like Pollen, they can reach us on my LinkedIn page. Any of those channels are okay and we are pretty active on different events and conferences. We are trying to be present. So I hope I can meet many of the audience in person.

Peggy: Well, I'll certainly link in with you and continue the conversation there. I have to say, Oz, it's always fascinating to hear about creative optimization. That is top of mind.

I'll be watching what you're doing and I may have you back to talk about it again. But in the meantime, thanks of course for being on the show, thanks for sharing.

Oz: Thank you. Nice to chat with you.

Peggy: And of course if you're a UA marketer out there. Someone who has superpowers, someone who maybe helps gaming companies literally up their game, either side of that equation, then hook up with me or with Pollen VC on Twitter or LinkedIn and we'll get you set up with a show of your own.

V/O: Stay up to date with our show schedule by signing up for alerts at pollen.vc. That's where you'll also find our suite of financial modeling tools to help you plan and manage your business growth.