Tabarak Paracha
Our CEO, Christopher Farm, appeared on the ‘GameMakers’ podcast to discuss our client HyperBeard’s success and Tenjin’s role in it. The talk focused on Portfolio LTV, a key analysis for reducing user acquisition costs.
What does a MMP do, and how has its role changed since the deprecation of IDFA?
[Christopher Farm]: The role of an MMP historically was to figure out which sources of traffic that you are advertising on produce the highest ROI. I think that is still the case but it has changed a lot since IDFA deprecation and moving to a world where things are less transparent. So one of the things that we differentiate on (at Tenjin), is that we try to pull out all of the data in a single place. We provide a data warehouse as a service, so that companies don’t have to build their own data warehouses. This is one of the ways we are kind of different from a lot of the other MMPs. And as a result, even with the deprecation of the IDFA, our belief still remains that you can pull all the different types of data that you can get. It’s arguably even more necessary that you do that and have partners that can analyze and look at that data appropriately.
What is Portfolio LTV?
[Christopher Farm]: The notion of Portfolio LTV – I don’t think it was something we were looking to do from the very beginning. It was this natural thing to look at when we learned what the customers wanted to do, or in this case Alex wanted to do, with the data.
We started to see from the data that it was very natural for customers to want to play similar games from your portfolio of games. And so it kind of naturally generated from that as an idea and turned into a lot of different analyses that we tried to do, and kind of optimize that experience for the customer.
I think it’s in the very early stages of the industry where people can think in a sophisticated way about this. There’s a lot of things and opportunities where I can personally see that if you can create a branding around your portfolio, you have a real shot of doing something like this, and there are a lot of product ideas that can spawn from this. And I don’t think people have even tried this, but having for example a universal currency, or very similar experiences, or sharing different types of virtual goods across your apps. All of this becomes a lot more possible with optimization and when metrics are there.
There’s an opportunity cost there, in the sense that you could have shown another ad of another publisher, but it’s a much lower opportunity cost and you can build your LTV this way. And it’s actually one of the ways that the math of hyper casual publishers existed. They would be acquiring users for much more than they could make in any single app. But taken as a whole, it was generally profitable for them.
When you guys were evaluating Tenjin, was Portfolio LTV one of the primary use cases you were thinking about or was that a nice to have?
[Alex Kozachenko]: We were thinking about Portfolio LTV before (starting to work with Tenjin), but it was still something that was a step away. One of the things that Tenjin did offer was custom dev services in the contract that we have, and that would allow us to sort of build the BI. So we knew that was the case, and that was the vision I had at that time.
The other thing that Tenjin allowed us to do was to expand our UA pool. The thing about having low LTVs is that you really want to have low overhead on top of your costs. And ultimately most MMPs are charging you on a per install basis tracked back to a UA source, or any sort of tracking source. So some of the bigger MMPs out there are not especially cheap. So if you want to expand outside of tier 1 – goodluck. It depends on the scale obviously. So if you have massive scale, you can negotiate a deal. But we can’t pay 2 cents for an install in India where the LTV is 1.5 cents for example. It makes absolutely no sense. So this allowed us to open up the pool. And right now we are seeing a lot of success where our global campaigns are running, which we wouldn’t even have thought about running on an MMP where the cost structure didn’t make sense.
Could you dive into the case study that Sonamine did with Hyperbeard and also talk about some of the services that you provide?
So we did a lot of the initial exploration to help him out a little bit. And we discovered a lot of opportunities that allowed Alex and his teams to strategize and prioritize where they would be putting some of their UA efforts. One of the reasons that we were called was that when you look at the traditional MMP’s when you run a campaign, you get the cost and then you also get the revenues associated with that campaign. But they tend to be tied to just that one game. But there is no easy way to actually see the Portfolio LTV for that campaign. That’s where we were able to help Alex and his team figure out a different way to do it, and to see it so they can use it every day to figure out the revenues for each campaign.
What kind of data analysis did you help Hyperbeard with?
[Alex Kozachenko]: So Tenjin and Growth FullStack were the tools that we used to bring the data in. But obviously we needed to clean it up and make it serviceable, which means creating the right dashboards, the right reports. One of the things they did was a custom analysis which let us know we have X% of users moving between 2 games, Y% moving between 3, and Z% between 4. This is your largest feeder game, the one that users are coming in through. This is the one that they go to most likely, and here’s the next one, here’s the next one. And some of the things we can do with that is we can structure our cross promo such that we show ads in those proportions based on where users are likely to go, and also where they’re likely to monetize. Because the whole idea of the Whiteout survival question is you want users to go where they’ll give you the highest LTV. But obviously you need a high probability that they’re going to go there otherwise you’re just showing an ad for the sake of showing an ad. So those are some of the use cases that they’ve helped us with. And like I said, because we don’t have any data analysis in-house, it’s been really helpful to have their help. [Nick Lim]: And we think of it as data is sort of a journey. So when you start out and you have a game and you are running UA campaigns, we can help on that front. If the data is present, we can generate SKAN prediction models, for example. Where, even though there is IDFA deprecation, we can still with the conversion values that are coming in, predict to within 5% of accuracy what the LTV would be like based on the conversion values that are coming back for that particular SKAN campaign. But sometimes we realize that the SKAN conversion value schemas are actually restricting what you can predict so we’ll go back and tell the developers they should use a bigger and broader SKAN schema that can capture more conversion values, not just revenue. That’s the go to for most developers today – to just put revenues in their SKAN values. We actually recommend that you don’t do that just because you’ve lost a lot of data already. So we will help design the SKAN schema to get you better predictions at the SKAN level if that’s something you want to do. That’s on the UA side.Sonamine also helps on the monetization and the retention side. So the minute a user starts playing a game, that’s only half of the challenge. How do you keep them, and make them spend money when they feel like they’re getting value from it? So we run AI predictions on top of the data that you have, and we identify the segments of users that are going to leave and we help you nudge them along if they’re going to convert. And if they’re going to leave, we help you provide some rewards that might help to keep them.
Watch the full video above to learn How to calculate Portfolio LTV, Big upcoming trends and changes coming up, and more.
