Case Study

The adoption of in-app bidding is gradually picking up pace in the mobile gaming industry, as leading developers have started integrating it into their monetization strategy.

One such company, Jam City, turned to ironSource’s in-app bidding solution, LevelPlay, for their hit game, Panda Pop! ironSource caught up with their monetization manager, Kiel LeBaron, to hear about its impact on their business. 

Time is priceless

Jam City saw LevelPlay as an opportunity to automate the time-consuming manual management of waterfalls across various geos and apps, each with dozens of instances. Saving this time enabled Jam City’s monetization team to shift their focus onto the product, communicating closely with the product team to optimize the placement of their ads. "That time is really maximized and could theoretically be a 50-100% to 2x increase in overall ad revenue," Kiel told us. 

Improving KPIs

One of the main selling points of in-app bidding is increasing competition for every impression. Jam City saw this in action: using LevelPlay, they improved the performance of key metrics, leading to both increased revenue and a better user experience. “We saw an increase in impressions and fill rate through rewarded video, not banners. That was a surprise to us. We also saw our retention increase quite a little bit," Kiel explained.

There was a positive correlation between rewarded video ads and retention: LevelPlay freed up time for Jam City to optimize the placements of the ads and, in turn, retention. 

Key tips 

In-app bidding may be an automated process, but there are certainly ways to optimize its performance, as Kiel explains:

“I’d say from a key tips perspective, definitely build towards an A/B testing solution. I think that’s first and foremost. The other tip is have really strong dialogue with account managers at the respective networks that are in and competing in automated waterfall with bidding because they might have some insight that you might not be able to get from the front end”.

Watch the interview below or read the full transcript to learn more insights into in-app bidding from Kiel and the impact of LevelPlay on Jam City’s business.


Reuben: Hi Kiel! As a long term mediation partner with ironSource, we want to give our audience insights into how a super successful company, like Jam City, approaches in-app bidding. 

The transition to in-app bidding has been catalyzed in part by the limitations of waterfalls. What issues would an ideal in-app bidding solution solve? 

Kiel: For in-app bidding and what we hope to achieve there, assuming all partners can adopt the new technology, is a more level playing field where we want everyone to be able to compete in the same auction regardless of rank in the waterfall. A lot of what’s been happening over the last couple years is publishers have gotten more sophisticated with their waterfall: more price points, more gaps in the waterfall, increased latency.

What we hope in-app bidding will do is flatten that waterfall and allow any network, or partner, or DSP to be able to bid on the inventory at a rate that they think is fair. Right now, we are playing games between what rank you are in the waterfall, and it’s blind to a lot of the partners. In-app bidding provides us a solution that makes it fair for everyone to compete when they see the request coming in. 

Reuben: At first, a lot of people in the gaming industry were skeptical about in-app bidding and the impact that it would have. Were you also skeptical at first? 

Kiel: I actually wasn’t. I have embraced the technology. I think it’s been on a desktop for a while now and, for whatever reason, it’s taken a long time to get into mobile. I think we’ve dug this in, a little bit, in terms of traditional waterfall, and it’s been hard to adapt. It takes a huge momentous shift to go over to some new technology like in-app bidding, but I think the future has already been paved on desktop.

Mobile does things a lot differently, but I think in-app bidding is tech that we should have brought over and continues to be developed right now. So, I’m excited for it. I think the skepticism was more on the advertiser and demand side. Advertisers saw that, on a desktop, there was a 30% price premium.

Obviously, on the publishing side, ad monetization teams loved that, we love to see our rates increased 30%. I think the reality is somewhere in between. There is a good chance that demand gets more efficient. That buyers are really able to see, once they are able to understand the request and, potentially, what the full waterfall looks like and a first price solution, that they are able to bid more effectively and efficiently on that audience. There’s a chance that they get reduced prices. Either way, I think it’s a little more programmatic and automatic with quicker responses. We get ads responded quicker for users and, in the end, I think that’s the best solution for where we sit in adtech on mobile. 

Reuben: We are seeing some encouraging numbers for publishers using ironSource’s LevelPlay solution. Could you talk about your monetization strategy and the impacts of LevelPlay?

Kiel: At Jam City, we want to treat the players first and foremost. We really care about their experiences. That comes down to ad quality and what type of ads they’re seeing, as well as whether or not they’re able to receive an ad and whether or not they’re able to get the reward. Obviously, once that’s accomplished we want to make sure the performance is there.

What we’ve seen with LevelPlay, in terms of performance, has been strong. I think we’ve got a fairly robust waterfall as it stands. We do a lot of yield optimization on our traditional waterfall. We actually weren’t expecting ARPDAU increases or CPM increases, we just wanted revenue to be flat, hopefully. A slight bump would be, obviously, fantastic. What we saw was an increase in impressions and an increase in fill rate and this is through rewarded video, mainly, not banners. That was a surprise to us.

We also saw our retention increase quite a little bit. We haven’t really understood why that is, but the impression increase is a good sign. Theoretically, with a reduction in latency, we should see an increase in impressions and that does seem to be the case through A/B testing through LevelPlay. We are excited about that. We’re going to be rolling it out on more games in a testing environment just to make sure the numbers are holding true on additional games of ours. 

Reuben: Since adopting in-add bidding, have you noticed changes in other parts of the business?

Kiel: Part of the goal, again, with the overall move to LevelPlay or bidding, is a reduction in operation costs. We spend a lot of time looking at yield and doing yield optimization. Theoretically, LevelPlay just automates a lot of that so we don’t need to spend a lot of time managing ranks in the waterfall, working back and forth with the networks on optimal price points, creating new, etc. That is a huge time sink for a lot of our teams.

What those teammates are now doing is going back to the game developers and working with our product teams to figure out optimal placements for ads in the games and looking at how users are interacting with our ads. That is time that is really maximized and could, theoretically, be a 50 to 100% to 2x increase in overall ad revenue if we find placements that our users are resonating with and is bringing in incremental ARPDAU.

It’s not that we’ve automated their work flow, it’s more that we’ve shifted the focus to more on product perspective which is fantastic and, ultimately, that’s something that is Jam City’s vision with regards to how the user environment is with games. 

Reuben: And that’s something that was not really possible beforehand with waterfalls? 

Kiel: It was just an additional time commitment. We’re looking to hire people specifically focused on yield optimization and to manage the waterfall. Waterfall management does take a lot of time. The big drawbacks are the back and forth with networks, obviously, the analysis behind it, and whether or not price points are filling.

If you do geo-optimization, that’s another huge investment, and not always is the reward the juice worth the squeeze, so to speak. Bidding just automates that. We’re not worried about it. Especially in a first priced auction, we are pretty much guaranteed the best rate for that user. Assuming the competition is there, that we have a lot of demand partners competing, we should get the highest price possible. That is something that is scalable as well because we’re not going to be able to set price points in the waterfall to infinity, right?

Ultimately, LevelPlay and bidding is changing how we do yield optimization because, traditionally, in yield optimization in traditional waterfall we have to guess what the price is from the network and the advertisers, we don’t really know. I call it a game of battleship where we’re trying to hunt and find the optimal price point, but we don’t know and the networks don’t really know that either. It isn’t until we’ve created that price point, look at the fill rate, and move things around that we’re able to see what that price point is, and those fluctuate. That’s the game of yield optimization, and LevelPlay, again, automates that so we don’t have to go hunting. The advertiser gives us that price point that they’re able to pay for that user, and that is the most efficient way that I think we can hold an auction. 

Reuben: Apart from yield optimization, what do you love most about LevelPlay?

Kiel: I think LevelPlay is fantastic from a product perspective, I think the front end is really nice and clean. I think when you combine that with the A/B testing tool that ironSource has, it’s really powerful, especially, as developers get more familiar and comfortable with how bidding works and how LevelPlay works.

The A/B framework gives you a very clear read of how bidding is doing compared to a traditional waterfall. To be frank, LevelPlay, bidding and the industry is really kind of new. A lot of networks are getting familiar with it, a lot of the mediation layers are getting familiar with the technology, and there will be some hitches. I think the visibility ironSource gives with LevelPlay also provides good transparency if there are issues to help both ironSource and the game developer go in a QA. To summarize, I think LevelPlay, when combined with the A/B testing framework, is really really powerful. 

Reuben: We touched on it earlier about the RV results that you saw. Are there specific ad formats for which in-app bidding is particularly effective?

Kiel: Jam City has been really focused on rewarded video. We feel that’s a win-win-win for the advertisers, the users, and, us, at Jam City. Most of our product portfolio is based around rewarded video; however, we have done testing with banners and banners are what see a major impression increase because of reduction in latency.

We also see an impression increase on rewarded video and we believe this is because of specific placements within the game. If users get the opportunity to see rewarded video or intersitials, for that matter, early on in the game, they might have this issue with latency. Because the surfacing point is so soon on, it might not have enough time for the auctions and SDKs to initialize in the background. That’s the theory behind why we’re seeing an impression increase on rewarded video.

In terms of the retention, we’re not really sure what’s causing the increase in retention. We know, generally, that users that watch rewarded video have an increase in retention. We’re hoping that is translated to the increased numbers there. Overall, impression increase is great, especially internationally. I spoke with someone else at ironSource a couple weeks ago about our strategies around yield optimization internationally. We don’t focus on specific geo regions, we just do a global international waterfall and that might have negative effects internationally. LevelPlay and header bidding solved that, as well, for us. 

Reuben: Let’s talk about key tips. You mentioned earlier, A/B testing. Obviously while in-app bidding is an automated process, there are things you can do to make it work better for you. Apart from A/B testing, what other things do you do to get the most out of it? 

Kiel: I’d say it’s getting familiar with the A/B testing tool that ironSource has. If you’re not using ironSource, definitely use your own A/B testing framework because it is crucial to understand what the tradeoffs are. We were surprised by some of the results, specifically in impressions, but there will be some surprises when you do the testing. And again, because it’s newer technology, we’re not really sure where things could improve or could potentially be just a fraction off from being a great success.

You want to be able to get as much insight as possible and have a control group to measure against, which is the traditional waterfall. I’d say from a key tips perspective, definitely build towards an A/B testing solution. I think that’s first and foremost.

The other tip is have really strong dialogue with account managers at the respective networks that are in and competing in automated waterfall with bidding because they might have some insight that you might not be able to get from a front end. Having that dialogue is key. They might want to test different things, as well, between bidding and the traditional waterfall.

From a metric standpoint, what you want to look at is comparative metrics, such as, ARPDAU and impressions per DAU. Those are the defining metrics to see if bidding is overall having an improvement to the existing set of users. I think those are great starting points.

Also, setting the right expectations. If you go in with bidding and you’re expecting a 30% increase in the waterfall, that might not be fair given the technology. We have bidders and demand partners that still need to come on to solidify the bidding experience. Right now, hope for a reduction in yield optimization, hope that you’re getting close to the best price possible, and go back and revert to focusing on product and experiences with games and ads. That combined, I think, is pretty powerful. 

Reuben: What do you think is the worst thing a monetization manager can do with in-app bidding? 

Kiel: I’d say right now just don’t turn on one bidding partner and hope that’s going to solve all the problems. I think we’re still in the hybrid solution, right now, where some networks are not ready to bid in that environment. Pluck the partners that are in the traditional waterfalls, but support bidding, and transition those guys over rather than turn everything to bidding and do away with any network that is in a traditional waterfall. There is still a lot of value to the partners that are in the traditional waterfall and I’m sure they are working on their bidding solution right now. That is more timing related than anything, but you want to give them a chance to compete in the auction.

Demand is great. Think eBay. If you only have one or two buyers, you’re not going to get the best price for your item. You want to have a competitive waterfall. That’s really a mistake I see some people doing, is just flipping the switch to bidding. I think that’s bad, not only from limiting the amount of demand partners but also monitoring it. Obviously, you want to check in on these metrics, obviously you want to look at things like ARPDAU and impressions per DAU, engagement rate, theoretically. All of these are metrics that are available on ironSource’s front end, very easy to access. It gives you a good insight into what is actually happening within the bidding environment. 

Reuben: What’s the biggest screw up you’ve made with in-app bidding, if any?

Kiel: That is a good question. I’ve made mistakes, definitely, with bidding, and it’s almost ad trafficking 101 back to my early days of ad operations. Make sure those IDs are correct. When you’re setting things up, you’re gonna be juggling a lot of IDs. Especially, if you’re doing an A/B test, you have to create brand new IDs to be able to isolate the tracking. Get that google doc going or spreadsheet going that really lays things out so it’s as easy as copy and paste, and double check your work. I’ve made that mistake and it can ruin an A/B test. Spend the double amount of time that you think is necessary to go in and double check your work and it will save you a lot of headaches in the future. 

Let's put these tips to good use

Grow your app business with ironSource