A look into the future of DSP for UA

February 12, 2020

Want a look into the 2020 outlook for mobile UA on DSP? Buckle up! Goodgame Studios is an entertainment software studio that designs and develops online games, and Pau Quevedo is the company's Lead of Programmatic Trading. Our Host, Peggy Anne Salz, caught up with Paul to talk about why you should be running DSP for mobile UA, the trials and tribulations of inhousing DSP, and what inventory works best for mobile gaming.

 

Peggy: Hello and welcome to "Mobile Growth," the podcast series where frontline growth marketing experts share their insights and experiences so you can become a better mobile marketer. So, I'm your host, Peggy Anne Salz, from Mobile Group where I help my clients grow the revenues and audience for content marketing, and I'm collaborating closely with "Mobile Growth" for this valuable actionable series and on my watch, that's what we're going to do. We're going to introduce you to the people who know how to drive growth, true to the name "Mobile Growth." And we've been going through a lot of episodes, I encourage you to check them out over at mobilegrowthsummit.com. Our guest today, we're going to be talking about programmatic, and who better than Pau Quevedo. he is lead programmatic trading at Goodgame Studios, part of the Stillfront Group. Hey, first of all, Pau, great to have you here, how are you doing today?

 

Pau: Hi, Peggy. Good to be here, all excited. Thanks for inviting me.

 

Peggy: And you can be very excited, actually, because, you know, hey, you've got some good news over there at Goodgame Studios. What's going on there? I mean, I say we talk about growth, but hey, you're living it.

 

Pau: Yeah. And Goodgame Studios, we are part of a Swedish group called Stillfront, which is one of the leading gaming groups for free to play. And this week, we announced that the acquisition of another company joining our family is called Storm8. They're from the U.S. and it was acquired for around $300 million as the press release says. This company is mostly focused on smash-up and casual games, which it's great for us because it opens up new genres. Stillfront has been focused on mid-core gaming and this is a different type of games that Storm8 develops, which opens up for a new audience and new deals and we're really excited about it.

 

Peggy: Yeah, now so as you said, it gives you that breadth because now you've got the casual, and we all know what the reports are telling us about the casual games and that market. I did some research and mid-core is also in a great place because people who play mid-core, they're, generally speaking, very loyal to that, more loyal than say, for example, hyper-casual, which is usually a retention curve of about, you know, 14 days if you're lucky. So, it's a good place to be. And of course, you're working in programmatic trading, you're a lead there, and your focus is, of course, bringing programmatic activities in-house. I'm just curious, what's the update on that? How's it going over there at Goodgame Studios?

 

Pau: Because I recently joined, I came from another gaming company here also based in Hamburg, and we are building up the team. We are going to test several tools, several DSPs that we find. We also have browser games and mobile games but our focus is at the moment mobile games. We are using completely different DSPs for mobile and for desktop. And the idea is mostly that for desktop, we focus on triple-A DSPs but for mobile, we need mobile-first DSPs or let's say not omnichannel DSPs, which are more specialized into the mobile gaming ecosystem, so to say, because it works completely different from desktop. So, at the moment, we are currently running a couple of managed services, the ones that everybody knows more or less, and we're trying to in-house new mobile DSPs. We're currently in-housing one of them and yeah, and it looks terrific at the moment. We don't have any results yet because we're starting right now but we're really looking forward. Our idea is to in-house it as much as possible with the long-term perspective of actually building our own custom algorithm, and that's the idea behind it.

 

Peggy: So, what's it like in-housing a DSP? I mean, first, you have to choose, I mean, you said mobile-first, you know, app-first, even better. You know, some of our listeners, they're all like you, you know, practitioners, growth marketers, in the business. What does that list look like these days of the candidates you can even consider for in-housing?

 

Pau: Well, that's a pretty good question because it's one of the toughest parts of this whole business is to actually choose the right tool, right? I would say the architect landscape is pretty much infinite, there's basically a new DSP every week, seems like. So, what we are looking at is we try to get the DSPs that they have the most advanced algorithms. That sounds like a lot, and obviously, when you're scanning the DSPs, you cannot really tell who has the best algorithm, you actually have to test them out. And for that, what we normally try to do is we try to find out if the algorithms that the DSPs actually work with are based on user-level data. That's the first indicator that things should be fine because there's a lot of DSPs out there that have very simplified technology, but that doesn't really cover our needs. And then apart from once we believe that the algorithm or the bidder is reasonably good, we try to find out if the QPS that is the traffic, the inventory that they actually listen to is aligned with our gaming inventory that we actually want to buy. Normally, that's the case, but we've had some surprises with some...especially with the omnichannel DSPs. They're not even connected to the reward video networks, for instance.

 

Peggy: So, that's another point, you know, indirectly another point of advice is, you know, not only do they have algorithms that are, you know, working and effective, but also do they have the partnerships that are effective? You said, for example, you have DSPs, you look at them and they're not really connected at all with the reward video network. So, that's a red flag right there.

 

Pau: Absolutely. Some of them are connected through base switch, which sounds okay, but we rather have them connected directly to the video ad networks. There's also a bit of problem with the video ad networks since they...you could say that they backfill the DSPs and within how they sell is in the waterfall system and the DSPs are normally penalized into it. So, that's also something we have to watch out. We definitely want to have the traffic from the video ad networks. But we also would like to have traffic outside the video ad networks for us, let's say, to develop more smarter strategies in the way we acquire users. And that is to when we actually work with user-level data, we are actually able to develop what we call the user ID graph that is to follow the users precisely. And we might identify those users in video ad networks but since the CPNs are so expensive, as you mentioned, hyper-casual is quite present in that traffic right now, we try to find them in the video ad networks but then acquire them as well where the CPNs are way lower. That would be a simple strategy to put it, how we use the user-level data in order to optimize our campaigns.

 

Peggy: That's also a very smart one. I haven't heard that. I've been in a lot of conferences, you know, I haven't heard about, you know, seeing them in one network but acquiring them, of course, where it's cheaper, where it's better in another. Is that a difficult process or is that something now that you and your colleagues out there, other growth marketers, are employing? Because it makes perfect sense and also, you know, in the age when we're really thinking about sort of multi-touch and getting people in through different channels, and now we can understand and track that better, you know, it makes good business sense.

 

Pau: Absolutely. It's quite tough to do that, honestly. Only DSPs are actually able to...the data is based on user-level data can actually pull that off but even in that case, it's not easy at all. I've talked with very, let's say, advanced DSPs and I've actually tried to make this happen, but the truth is that to create the user ID graph, not every DSP can and you need a very, very heavy investment in data framework systems and also in machine learning, let's say. We're not there yet, but that's exactly where we want to get to in order to understand the data that we have at a user level and derive simple strategies that you would normally think even in supermarket, I don't know, whatever, and use those techniques in how we acquire users in the online marketing space.

 

Peggy: And using that user graph, I would imagine also attacks and other problem hearing...well, not a problem but a question now on everyone's mind, which is, you know, how do I keep incrementality in check? You know, money, cheap money...era of cheap money is over, and you sort of want to not only know where can I acquire them cheaply but can I really see the user through the journey? I don't want to acquire them again, you know, pay twice or when they, you know, reinstall, you know, that I don't call that as a new install, I'm calling that you know, a returning user. I mean, are these the kinds of questions you're grappling with over there?

 

Pau: Yeah, exactly. The cool thing about DSP is that we try to keep everything under one roof. So, that allows us to actually understand that process much better. The issue is that, and the realities that these teams normally, at least the gaming companies I'm familiar with, we are way smaller than the Facebook teams or UAC investment, let's say. So, it's quite hard for us and I mean, since those work, like walled gardens, black box, it's very hard for us to really determine the impact. And also what you mentioned about the incrementality and the organic, if we're actually eating organic, let's say, that's something that we constantly check, but it's not that easy to tell. That's honestly so.

 

Peggy: Well, thanks for that Pau, I mean, it's great to have a straightforward conversation. A lot of people are thinking about DSPs, going programmatic in 2020, exploding market, really hot topic. We do have to go to break now, however, so don't go away. We'll be right back.

 

Man: If you haven't already, be sure to check out the Mobile Growth Global Event Series where thousands of the world's top growth marketing experts come together to discover the latest in growth marketing innovation. We run events all around the world in cities like San Francisco, Los Angeles, New York, Berlin, Montreal, and more. And our candid panel discussions feature industry leaders including Facebook, Google, Uber, EA, and loads of other top tier publishers. So, attend our lineup of intimate workshops and gain the in-demand skills you need to stay on top of your growth marketing game. Expand your professional network and build meaningful relationships in a fun, friendly, publisher-first environment. When and where can you get in on the action? Well, just visit mobilegrowthsummit.com for a complete list of upcoming shows. And if you should decide to join us, and I hope that you will, then be sure to use our special promo code MGSPODCAST30 for an additional 30% off your ticket order. Once again, that's MGSPODCAST30, all caps, no spaces for 30% off of your order. We hope to see you there. I certainly hope to see you there, and please enjoy the rest of our episode.

 

Peggy: And we're back to "Mobile Growth." We have Pau Quevedo, he is lead programmatic trading at Goodgame Studios. Pau, before the break, we were going through, you know, just very simply, very calmly...that's what I like about you, you're very calm about an extremely complex subject that people really do stress about. You know, choosing the right DSP, how do I in-house, what's the best...and how do I test the algorithm? A lot of questions here. And I understand you may be on the lineup for one of our speakers at one of the Mobile Growth events. So, I really do hope to hear you there. Have you spoken very much?

 

Pau: Not that much. I've spoken in a couple of events since...yes. But I'm looking forward to speaking many more since I think programmatic is a topic that is growing quite a lot and mobile UA in programmatic, I mean, it just has to grow. At the moment, it's a bit small, but we see a lot of potential into it.

 

Peggy: I absolutely agree, and I do see it growing, I do see a lot of conversation about this. So, picking up on the idea we talked about, you know, what to look for, obviously, a mobile-first or app-first DSP, make certain they have user-level data, super important. What about your side of the equation? You know, you want to own the data as well, and you want to improve the algorithm, you have to teach the algorithm. So, it's a bit of a trade-off. How are you approaching that?

 

Pau: Well, it's quite tough, to be honest. The way we do it is that first, we have to also teach our marketeers how to actually understand the value of the data, which is not that obvious sometimes. And also the data scientists, they have to get an understanding of what we actually want here. And to get like a marketing mindset into the data scientist, that is not easy at all. And that's one of the first steps that we're trying to do. Then comes a lot of technical challenges from GDPR to expensive tech that you actually need to model all the data. So, we companies, we first have to prove the concept in, let's say, working closely with DSPs, we might be able to practice and make it happen and then convince, explain to the management that we see a huge potential. The great potential that I see into in-housing DSP is that currently, the DSPs they're focused on local-level data, that is the data that we get from the auctions. They use that data to teach their algorithms but they're missing a big part of the equation, which is what happens after the install?

 

Obviously, we send them events, we push them through the MPs but that's not really the same as all the information that we have in-house, what we call our Hadoop Data API. So, we see that the moment that we can actually merge the data that we have in our systems, API data, which we actually don't share with Facebook or any other partner, and we merge it with the local-level data, that could bring us such an advantage over the networks, over other different partners, but it's just a matter of time until the industry realizes and it shifts always towards that. We've also seen this and DSPs actually have understood this and we see how DSPs actually are already working with user-level data, that unfortunately, not all of them do. Those who do, they see that the next step in gathering data and being able to improve to grow, as you say, is basically to get more data. Where are they going to get it from? What happens after the install.

 

So, that's why we also see some DSPs turning into some sort of SaaS company business model, which what they try is to ingest all the data as much as possible from the advertiser, from us, in order to be better predicting the bits, be better predicting the strategies, be better are growing in an online marketing ecosystem. And convincing, making that happen is the biggest challenge that we have. And that is something that is not going to be only us advertisers, we need the help of the DSPs. I see it as a joint business myself. And because I know some gaming companies have already tried to build their own bidder to be independent from the DSPs but that doesn't...it hasn't worked out. We realized that it's extremely difficult to build all that [inaudible 00:15:15]. So, it's a combination of both. But I think it's very important that marketeers, we realize the importance of data is really how we use it and how we train the algorithms with it. And I think all these cuts in the next couple of years are going to circle around how we actually train the algorithms, what data is relevant, what it is not and this kind of thing.

 

Peggy: Is there any concern there? Because I understand that, you know, when you're talking to a walled garden, you're thinking about a Facebook. I remember being at conferences where people are like, "I don't want to do this because I have a fear...you know, off the record," that's what they told me, you know, "there's the fear that I'm making them smarter and then they're going to turn around and make my competition equally smarter," right? So, I could understand that concern. It seems to be less in the DSP world. But is there a trade-off to be aware of there? Because I mean, you teach the algorithm, it gets really good and then your rival comes up and benefits from, you know, that level of intelligence. I mean, that's just the way it is, but how do you balance that?

 

Pau: Absolutely, and I think you really nailed it. There are two elements into this that are extremely important. One is that the problem that we're doing the advertisers is that when we launch campaigns, we're not just paying for the campaigns that we're actually doing, we're actually paying the algorithms to learn for us. That means I am giving the DSPs money for them to learn to use those learnings for other advertisers, which is okay because in the perspective face of the campaigns at the beginning, we want to benefit from what the other advertisers have learned. If you start off with your own data only, it's going to be almost impossible. So, that trade-off is okay.

 

The fear that I see is that right now we are working in an environment where the video ad networks, they work in some sort of oligopoly, which they have a lot of market power. And I fear that in five years' time, the DSPs have learned so much from all the data that we've been feeding them for all these five years, that is going to be impossible for us to then compete against the DSPs, that is to actually in-house the problem. So, in a way, it's very important for us that the DSPs get data, but in a way, we're shooting our own foot because, in five years' time or something, we will not be able to compete with the DSPs. And, okay, that's fine, but we are basically feeding another beast that will come.

 

There's also another problem when we talk about this data flows and how we feed the algorithms. And that is that right now, our work is in a waterfall system, right? So, you're used to getting some sort of data, but the moment that the change actually changes you in the waterfall system, you will be getting new data, which will pretty much make the data, your learnings are irrelevant because you have new data, which learnings cannot really be applied to. So, in any moment, the moment you make this huge investment of, okay, I'm going to try to model the local-level data, I'm going to try to do all these different things, there's a lot of risks that you could be easily hurt and disrupted. So, there's a lot of problems into this, this is things that we advertisers have to look into but sometimes we feel it's worth taking the risk.

 

That's also true that it's extremely expensive and that there are also even companies out there like Sybase, Ubiquity, other third-party companies actually provide algorithms or specialize into that. So, sometimes I feel that even if you put so much effort into in-housing certain programs, the third-party problem that another company comes and makes even a better product and cheaper is there, but that's something you can never control. So, those are elements that we constantly think about when we are thinking about how to in-house DSPs and the threats that we have.

 

Peggy: But it's great to compare notes with you because as I said, you know, I'm the analyst, I'm a mobile analyst over at Mobile Groove, I cover the industry for "Forbes," I know how it is from that angle. Your perspective as a practitioner is fascinating because then these are the concerns you have...no answers for them. If we had an answer, hell, you know, Pau, you'd be writing a book or you'd be on the circuit, talking and giving the answer. So, we can only really just think about that in terms of these are the top questions you have to have at the top of the business agenda, and we are in 2020. I'm just curious, looking out...year is just kicking off, questions on your agenda, you know, things that really concern you. we're talking about in-housing but, you know, is another item on your agenda, perhaps, also how to set up the data stack, if we want to call it that, that's going to make certain you get all the data to all the members of your team who need it and get everyone sort of on the same page? Because that's something I'm hearing a lot about too.

 

Pau: Yeah, we actually have some sort of DMP built. And we are...I mean, gaming companies, at the end, it's basically like a data company. So, we have all those tools, but we haven't really used them to create a smart strategy. And I think that's really the importance and the key. And the most difficult part of it is not just to have the data, it's to actually make good use out of it. And that's where we actually have to learn, also marketeers. And that, I think, the next couple of years will probably be our main focus, how to structure the data and how to actually gain insights out of it. But if we're talking about 2020, to me, the most important element that will probably shape this year, because it has to do with what we were talking about the algorithms, which is related as well to the data part, it's the header bidding topic for the mobile.

 

Honestly, that's what we are waiting for. We hope that that will kind of like break the oligopolies that the video ad networks have, the ACK networks. And we see that that could open up tons of possibilities for us to be more free and to, let's say, reduce noise. The problem that we have right now is that we might have everything in place, our data, we might have all our studies, everything is fine, but since..the market where we buy, we sometimes feel it's not true. There's a [inaudible 00:21:17] that is not completely transparent, it's not as we were hoping for it. We hope the head of bidding will reduce noise into the market, into the auction. And that would allow us to be more clever and to be more independent of what we're doing.

 

Sometimes I feel that I see products, I see some of DSPs that have an excellent product, excellent algorithm, they are talking what they're doing, but they didn't get clear that there's a problem when you go to the auction, there's other players. And I see that reducing that kind of noise, what the waterfall is causing, will eventually help all the other tools to develop. Because you can't really develop all these user ID graph and all these crazy customer algorithms if, at the end of the day, you have a waterfall system where someone else, without all that tech, can just beat you by just putting a number. So, we also have to democratize this a little bit if that makes sense.

 

Peggy: That's exactly what I'm hearing. I was at a show, actually, it was a Mobile Growth show last year and in London, and I had a panel about, you know, in-app header bidding, we've been talking about it now for, you know, intensely for two years, you know, what's holding it back, and it's transparency. It's number one, transparency. It's not all it was cracked up to be yet. I'm just curious, if you could have, you know, your...well, actually, you do, you're here on my show so you do have your podium, give some advice. What would fix it? Is it really just transparency or is there something that could be done that would bring us a giant step forward? Because transparency is like one of these vague goals, "Oh, they need to be more transparent." Well, okay, great. But what's the next step to that? What would you welcome? What should they be doing? Maybe they're listening in and say, "Hey, here's how you get your act together so that we can start seriously using this." What would it be?

 

Pau: Well, if we look back at history, what happened here is that the video ad network, the ACK networks have a very strong relationship with the publishers, it all circles around the publisher. Header bidding is not going to work until the publishers are decided that they want to monetize through header bidding. We see products like Max or Moba [SP], which are really pushing this and we are very grateful. But I also hear from the publishers that header bidding is not easy, that they actually see less return from it. Let's say you're a monetization manager, this week you're running your campaigns on a historical CPI, then you get a certain return. Next week, you try we header bidding, you have 5% less. Will you stop? Because this is what is actually happening. So, I think that it has to be someone has to suffer. Either we are willing to overpay for some time or the changes are willing to, I don't know, move this around that the publishers are better off with header bidding.

 

It's not until the publishers don't realize and understand the benefits of header bidding, the header bidding will not take off. Because they are comfortable in the situation with their historical CPIs that worked with CPI networks because it gives them a constant flow of cash. I know that the waterfall still needs quite a lot of time to adjust it, but the header bidding seems to be even worse. I was recently in London in UA Society and we saw an extremely interesting conference from a publisher explaining how they felt about header bidding. And you could feel that it's not easy for the publishers to implement. And I just hope that they come up with a solution that actually makes all the publishers overnight switch. And I believe that's the only way, money, at the end.

 

Peggy: That sounds about right. I think we were at the same event, UA Society, London, last year, September? No, October, October 4th.

 

Pau: Exactly, October.

 

Peggy: There you go. We almost met and hope to meet you also in person at an event out there somewhere, you know, Mobile Growth or other. In the meantime, Pau, how would our listeners stay in touch with you?

 

Pau: Well, they can look for me in LinkedIn or, yeah, in LinkedIn, basically, pauquevedo@goodgamestudios and I'll be glad to continue this conversation. In fact, a couple of years ago, when I started doing mobile DSP, I realized that one of the main problems that we had, at least in the gaming ecosystem, is that we didn't know much. I talked to other advertisers, and there was like a lack of knowledge, lack of understanding of what was going on. So, what we did is that we rounded up and we created some sort of programmatic roundtable. And we have our own Slack group where we only talk about these topics and it has been extremely good. What we do also is we try to meet twice a year, Moba is also hosting this event. So, we just need only advertisers and we just discuss these topics and go over it. I'm actually handling the roundtable and if any advertisers are willing to join, I think it's great because we actually share a lot of information and great insights over what's going on. We advertisers have to unite in order to understand the ecosystem. That's my idea behind...

 

Peggy: I support you wholeheartedly. I saw that coming out and I thought, oh, that's a great idea, you know, to really, really deep dive into programmatic. It's not what I would do, but I would be loving to understand it more deeply in order to write about it more effectively and offer more value. So, if you're taking non-gaming or non-app marketing members in, you know, please think of me too, I would welcome the opportunity.

 

Pau: Absolutely. Absolutely, Peggy. In fact, I think you know most of them who are already...you know a lot of them already. Like Misha from [inaudible 00:26:45]...

 

Peggy: Oh, yeah. Come on, Misha, he's also been on our show. I've done a podcast elsewhere with him. It just goes to show you, it's down really to community and sharing. I'm a huge fan of that. Sicolo [SP] who kicked off the UA Society, that was the whole idea. It was knowledge sharing and that's what I try to do here in collaborating and bringing the "Mobile Growth" podcast to everyone. So, we have run out of time but I am hoping that we will cross paths soon. Pau, it's been exciting to have you here and above all, so absolutely informative, so straightforward, you know. I'm probably going to take you offline and see also if I can't convince you at some level to write a couple of guest posts because what you know, you have to share, and I would encourage you to do that.

 

Pau: Thank you so much, Peggy. And I really appreciate that you gave me the opportunity to talk in your show. I was really looking forward and try to continue this offline, and hi to all your podcast listeners.

 

Peggy: Absolutely. It is a great community. I enjoy it, enjoyed having you here as well. And a quick reminder to everyone out there, hey, don't forget to visit mobilegrowthsummit.com for a complete list of upcoming events. Hopefully, one where Pau will be speaking, who knows what's in the stars there. And don't forget to use the very special promo code MGSPODCAST30 for 30% off of your offer. I hope to see you there. I also hope to see you over at mobilegrowth.com where you can find my portfolio of content marketing, app marketing services, and of course, all of the episodes of Mobile Growth Summit are on Soundcloud and coming soon to more channels making it possible for you to listen in, in more ways, and also to catch up. So, watch for that, we'll watch for you, and we'll see you soon.



Leave a Comment

MGS Custom Ad Spot

Want to be featured here? Submit a request below and we'll contact you.