ABOUT: LinkedIn  | Twitter | Kaizen Internet


ABOUT ROCKETSHIP HQ: Website | LinkedIn  | Twitter | YouTube

KEY HIGHLIGHTS

🤔 What is ad mediation?

🙋‍♂️ How mediation layers are connected to individual ad SDKs.

💡 Why prioritizing a waterfall algorithmically or programmatically can lead to suboptimal revenues.

📊 How eCPM floors work.

📞 How different calls in the waterfalls work.

🙆‍♂️ How many waterfalls an app should set up – and how an app should think about this.

👨‍🏫 How an ad monetization manager needs to find the right balance between fill rate and eCPM floors.

💵 Why ARPDAU is a more important metric than eCPM floor or fill rates.

🎮 How the depth of a game’s content pipeline defines its ad monetization strategy.

💰 How some games are built around rewarded videos.

🤷‍♂️ In-app ads can result in churn – and also incremental monetization: how developers should think about this.

🙅‍♂️ Why blacklisting competitors may not always be a good idea.

KEY QUOTES

What is ad mediation?

Ad mediation is basically a piece of software that mobile app developers integrate within their portfolio of games or apps. And it is used to manage and optimise all the ad inventory by creating a competitive environment where all the ad networks or supply partners fight, compete for impressions in order to serve the advertisers’ campaigns.

How ads should work on mobile

But the thing is that, the difference between a programmatic ad exchange and a mediation is how the ad impression is served. So we know programmatic ad exchange theoretically should have all the supply partners at the same level. And then the ad request by a user comes in so all the partners receive the request. So they say, do you want to bid for these users in this format, in this country, bla bla. And they say yes, this is my bid, this is my bid, this is my bid. So, the system takes these in real time, and the bidder is more profitable for the publisher. In the mobile app industry, it is not the case because nobody’s doing it right or hasn’t a perfect system. 

The clear case for multiple ad partners

It is recommended to have a wide variety of partners. There’s a lot of documents where they list who is the best for this format in these countries. So depending on where you have the traffic from, you should integrate one or another.

Why programmatic is often done manually

It could potentially be done algorithmically. The thing is that not all the partners are prepared for these algorithms or algorithmic buying. So, like I said, the only logic behind it is the technical capability of the supply sources. But it is important, now that we don’t have header bidding working fully. We need to take advantage of the waterfall because it has a direct impact on the revenues. 

How multiple ad partners impacts the bottom line

It is because it’s a way of forcing the supply partners to give you what you want. Because if you only have one partner, they can give you whatever they want, the first campaign, whatever. But if you force them to hit a target, I mean, they can or cannot, but if they hit it, it’s more revenue for you, usually than having it be automatic. And if you repeat this process with all the partners, you are maximising the revenue that you can get from a single ad impression.

The right metrics in the right circumstances

I would rather see the ARPDAU as a key metric, like Average Revenue Per Daily Active User as the key metric to look at, if we talk about ad monetization, because eCPM or fill rate per se, doesn’t mean anything. So, you need to maybe a 1% fill rate is; you could say it is low. But if you have 1% with an eCPM floor of 2 million, that could be a lot of money instead of having a 2% with an eCPM floor of 20. Because you will be getting more revenue. And the metric that you should look at is the ARPDAU instead of the fill rate and the eCPM.

Tie in your monetization strategy to your genre

So, if you have a game with a small content pipeline, the recommendation would be try to be as aggressive as possible with the ads, because otherwise you won’t monetize the users as well. Whereas other kinds—let’s say we take hyper casual versus an RPG or a real time strategy—hyper casuals, due to the simplicity of their mechanics and their content pipeline, they cannot afford not to show ads or not to be aggressive, because otherwise, their user acquisition won’t be profitable. 

A fine balance of intrusiveness

It is trying to analyse what the impact of the intrusive ad formats like interstitials and banners is – do they hurt retention or not? And what are they producing financially in exchange? So let’s say if you lost like 3% in d1 retention because you’re showing a lot of interstitials or your interstitial frequency is really high. But in exchange, your ARPDAU is 30% higher, you’re like, why wouldn’t you be even more aggressive or as aggressive but considering that you need to find the balance between losing retention and making more money.

FULL TRANSCRIPT BELOW:

Shamanth: I’m very excited to welcome Jorge Canga to the Mobile User Acquisition Show. Jorge, welcome to the show. 

Jorge: Thank you very much, Shamanth for having me. It’s been a long time since we first discussed it. And thankfully, we made it. 

Shamanth: We did indeed. And you’re certainly been recommended by quite a few people. So this has been exciting to have you. Also we spoke a while ago and I was definitely very, very impressed by the sheer breadth of work you’ve done. So I’m definitely excited to dive into one theme and topic that really, really is an area of particular expertise for you – ad monetization. We’re going to talk about how app developers can and should optimise their ad monetization stack. Lets just start off, can you talk to us about what is mediation? And why is this important to an app developer?

Jorge: Yes. So hopefully this may sound familiar to most of the app developers listening to the show. I hope it is more than that.

Ad mediation is basically a piece of software that mobile app developers integrate within their portfolio of games or apps. And it is used to manage and optimise all the ad inventory by creating a competitive environment where all the ad networks or supply partners fight, compete for impressions in order to serve the advertisers’ campaigns.

So this fact of competition creates, like an extra revenue for the publishers, because like in any kind of competitive market, when there’s more agents trying to like to get something, the price goes up.

Shamanth: Yeah, that makes sense. And this is perhaps a beginner question. As somebody who’s seen ad monetization from the outside, I’m curious: the way you describe mediation, how does it differ from an SDK? How does it differ from a programmatic exchange? 

Jorge: Okay, mediation could have a programmatic exchange on it. But the thing is that nowadays, programmatic has changed in mobile apps – it is not as developed as on the web. There’s like some people trying it, I think the industry is evolving towards the programmatic since it is like less hustle for marketers and app developers like to set up and it is said to have those results. 

But the thing is that, the difference between a programmatic ad exchange and a mediation is how the ad impression is served. So we know programmatic ad exchange theoretically should be like all the supply partners at the same level. And then the ad request by a user comes in so all the partners receive the request. So they say, do you want to bid for these users in this format, in this country, bla bla. And they say yes, this is my bid, this is my bid, this is my bid. So, the system takes these in real time, the bidder is more profitable for the publisher. In the mobile app industry it is not the case because nobody’s doing it right or hasn’t like a perfect system. 

So, right now, what normally publishers do is have all the supply partners integrated within the waterfall and they have, ordered like the different partners, depending on the country, depending on the ad format, and depending on the nature of the game, they have several parameters and they have it ordered, like on a waterfall. So first, second, third. And only, like when there’s a request from a user, the first partner gets the request. And if they don’t like it, that comes to the second and comes to the third. So it’s kind of like a waterfall.

Shamanth: Right. Right. And that’s why the term waterfall, that’s how it works. Thank you for explaining very simply how that’s different from an exchange. And to the other part of my question, how is this different from an SDK? 

Jorge: Well, both are an SDK because SDK is a Software Development Kit. So any kind of piece of code that you integrate within your app could be called an SDK or API, but normally it is through an SDK. So both programmatic ad exchange, or a programmatic environment like header bidding, could be through an SDK. On your ad mediation, it is from an SDK as well. The only thing is that within the ad mediation and the current situation, for most of the publishers, you have like adapters for this SDK where all the networks are integrated. So if you have IronSource mediation, and you want to integrate AppLovin, then do need to integrate like an extra piece of code for AppLovin to get into that mediation, and so forth for the rest of the networks.

Shamanth: Right. So if you have a mediation layer, you need to add an adapter, but you don’t necessarily need to add five or six SDKs, separately.

Jorge: No, you can actually choose,

it is recommended to have like a wide variety of partners. There’s a lot of documents where they list who is the best for this format in these countries. So depending on where you have the traffic from, you should integrate one or another.

I would say you need to integrate at least five six partners to have like these competitive environments. So if you have a mediation and only have two partners, it makes less sense. So you would need to integrate these adapters for all the networks that you want to plug in. 

Shamanth: Right. So you want to have one mediation layer, and five to six adapters, or five to six supply sources that’s a plug into the mediation layer so that the mediation layer can actually prioritise five or six ad networks, programmatic exchanges, depending on the geo, depending on the priorities, as may be set in the waterfall. 

Jorge: Exactly.

Shamanth: Yeah. And you briefly described what a waterfall is, and, you’re like, look, that’s how you prioritise, let’s just say different placements, that could be interstitials, rewarded video, whatnot, different countries, different monetization levels. Why is it important to manually prioritise this? Why can’t this be done algorithmically? Like, you know, you described how the programmatic world works. Why is it that it cannot be done algorithmically?

Jorge: Yeah, I mean,

it could potentially be done algorithmically. The thing is that not all the partners are prepared for these algorithm or algorithmic buying. So, like I said, the only logic behind it is the technical capability of the supply sources. But it is important, now that we don’t have header bidding like working fully. We need to take advantage of the waterfall because it has a direct impact on the revenues. 

The fact is you not only are selecting who comes first and second, you need to like to get a little bit deeper and try to like to set up some floors, meaning that target eCPM prices for each network that you can actually have several of those per ad network or per supply source, because it is proven to increase the revenues of the publishers just by like 30%. So like the main difference, or the main advantage of using another mediation and having a waterfall is that you can give like or see a 30% increase in the revenues just by setting another mediation with manual priorities. But if you want to get deeper into the ad monetization world, it is recommended to set eCPM targets. 

So, these eCPM targets are actually a target that you said to the ad network that they need to hit – otherwise the request goes to the second or third and so forth partners. So in this way, you can also get a plus 30%, plus 25% uplift compared to having a normal mediation with a manual ordering without any eCPM floors. That’s the kind of the smart way to get away with ad monetisation these days.

Shamanth: Right. So just so I understand if you’ve set an eCPM floor, you will get no impressions that are below this eCPM level. And, if you get a guarantee that, okay, let’s say you set an eCPM floor of 10, you get a guarantee that every impression you get is over $10, which is the advantage over just having one partner or without a mediation, you could get below 10, above 10. And the advantage of having a waterfall with the minimum eCPM is just getting guaranteed inventory.

Jorge: Yeah. Exactly. So,

it is because it’s a way of forcing the supply partners to give you what you want. Because if you only have one partner, they can give you whatever they want, the first campaign, whatever. But if you force them to hit a target, I mean, they can or cannot, but if they hit it, it’s more revenue for you, usually than having it be automatic. And if you repeat this process with all the partners, you are maximising the revenue that you can get from a single ad impression.

Shamanth: Right. And so you’re like, first place in the waterfall – is that the right terminology to use? 

Jorge: Yes, like the first call. 

Shamanth: The first call in the waterfall has the highest eCPM, you get X number of impressions, the second call you have a lower level of eCPM, and so on and so forth. And each call – is that for a separate placement, geo, how is that or is it just overall?

Jorge: Yeah, you must divide it between ad format, country and then I mean and operative system of course. So let’s say you have Candy Crush, iOS, rewarded video, United States. So you have a waterfall and then you have Candy Crush the same but instead of the United States you have India or Brazil, but you have a different setup for each country or group of countries if they have, kind of similar eCPM.

Shamanth: Right. How many waterfalls do you recommend developers set up? Or if that is the wrong question, how should they think about this?

Jorge: Yeah, like you can have as many as you want. But sometimes it becomes crazy while optimising because if you have like a lot of waterfalls, and you have a lot of ad formats, it becomes a mess. And if it’s like a one man team, that could be like a one app developer that is creating their games. They have no time for this stuff. I would just look in tiers, maybe the United States is separated and then the English speaking markets, then kind of tier 1 European markets, then Latin America together, then South East Asia together, then rich Asian countries or Japan plus Korea. So depending on where you have the traffic from. 

If most of your traffic is coming from, let’s say, Russia, then you should have a waterfall specifically for this country – and one waterfall for each country and ad format. So I think that’s the correct way. And also trying to add eCPM targets that we said before, it is a must if you want to get the most out of each ad impression. I mean, in bigger organisations where they have like a five person team of  ad monetization guys or maybe bigger publishers, it is obvious, this is happening with a level of accuracy that it’s really impressive. 

But, if we talk to medium sized app developers, it is not always the case. And there’s like people that they simply don’t care. But like, why wouldn’t you take, like an extra thousand bucks per month – if you can just optimise on a weekly basis, on a daily basis, depending on the time that you have. But at least if you set it up, then you have the warranty that at least you’re gonna make more money than if you just use one supply partner or just on mediation in autopilot.

Shamanth: Right. So it’s almost like you look at the volume of impressions, and you look at your target eCPM level at different geos. Then you set up okay, how many waterfalls do I need to have? And just so I understand let’s just say the first call, you set $10 eCPM for rewarded video, US. Does that have to be one single supply source or can multiple supply sources fill that inventory?

Jorge: Yeah, if it was like a programmatic thing, all would compete for the same but given that it’s a waterfall, you need to set up three eCPM targets per supply source. So, if you have just to make it simple, Facebook on its network, AdMob and Unity ads, you need to set like three eCPM targets for these three networks. So you will have, let’s say, maybe in the United States is Facebook audience network first with 25, then comes AdMob with 22, then comes Unity with 21, then again comes Facebook, then again comes AdMob then Unity and then the process goes over again. And then if nobody has made these requests, then it goes to the normal waterfall like the auto mode, which then decides who deserves this impression.

Shamanth: Right. So the one that’s bid at 25 gets the first opportunity to serve the impression, but it still can and if not, the second one does. Right? And the first person is called, that’s the first call in the waterfall, the second the second call, third is the third call. None of the three do, then it goes to the fourth call, but each call represents one network?

Jorge: One network and one eCPM target.

Shamanth: One network and one eCPM target. Okay, that makes sense and this is a hypothetical question. You describe, okay, for Facebook, you have a eCPM target of 25. And maybe AppLovin you have 23, what happens if AppLovin has an impression that comes in at 26? Does it get first bid? Yeah, what happens?

Jorge: Yeah, so that’s like the routine of the ad monetization manager. They need to analyse what is the fill rate and eCPM of each instance of the network. So if you say, let’s say in this case, so Facebook audience network has a fill rate of 1% at $25. But then the second call is AppLovin with $24. They fill 50% just to make it simple. I’m a smart ad monetization manager who would say okay, these guys are showing me that they have room for either dealing more or potentially making more money for me. 

So, what I would do in the testing round is compare how much I make if I put AppLovin first and then I put even a higher floor, let’s say to $27 or $30 or $35 or whatever the level that you want to set and then analyse what is the fill rate at that eCPM target. And that’s the way you optimise the waterfall, like at the end of the day, how much are you making per ad impression.

Shamanth: Right. So that’s an important trade off. Because you could set your eCPM floor at $100. But you probably get 50 impressions, so you’re just not making any money. But there’s always the trade off between impressions and the fill rate. Are there other metrics that you recommend, you know, an ad monetization manager look at on an ongoing basis to optimise a waterfall, to evaluate the health of the monetization stack?

Jorge: Yep. So first of all, I would not look into these parameters.

I would rather see the ARPDAU as a key metric, like Average Revenue Per Daily Active User as the key metric to look at, if we talk about ad monetization, because eCPM or fill rate per se, doesn’t mean anything. So, you need to maybe like a 1% fill rate is – you could say it is low. But if you have 1% with an eCPM floor of 2 million, yeah, could be a lot of money instead of having a 2% with an eCPM floor of 20. Because you will be getting more revenue. And the metric that you should look at is the ARPDAU instead of the fill rate and the eCPM

because people normally just say, “Okay, my eCPM is that, so I just pay attention to the eCPM and the fill rate but at the end of the day, what matters is the ARPDAU because it is tied to the LTV of your users. And that’s the metric that you should be looking at.

Shamanth: Definitely, I think that’s a useful reminder for people to keep their eye on the big picture, because it’s easy to be fixated on just eCPM or just the fill rate. Yeah. And how do the different ad placements play into this? Or one example I can think of is, if there’s an interstitial placement, that could be filled by a video or an interstitial banner. How do you recommend ad monetization managers think about prioritising different placements?

Jorge: That’s kind of a deep question. So, it depends on how the app or the game is designed. And ideally, the ads must be part of the core loop. So that means it could eventually add some value to the user experience. But it’ll be depending on the economics and the content pipeline of the games. Because that’s not always the case. Because like you said, there are not only rewarded videos, but also interstitials that are not decided by the users when they want to see these kinds of ads. So it is on developers to say when and how often they show interstitials.

So, if you have a game with a small content pipeline, the recommendation would be try to be as aggressive as possible with the ads, because otherwise you won’t monetize as good the users. Whereas other kinds of general, let’s say we take hyper casual versus an RPG or a real time strategy – hyper casuals, like due to the simplicity of their mechanics and their content pipeline, they cannot afford not to show ads or not to be aggressive, because otherwise, their user acquisition won’t be profitable. 

And on the other hand, like a real time strategy, or an RPG, or a game with a bigger content pipeline, where users can actually stay longer and day 30 retention could be higher, you don’t need to place interstitials or banners because they have a higher percentage of paying users, so they can buy in-app, but also include kind of rewarded videos or maybe an offer wall. So these are non intrusive ad formats, and it is only upon user consent, if they want to see an ad or not. So it depends pretty much on the genre and the retention metrics, just to try to make financially sense in the ecosystem.

Shamanth: Yeah. Right. And I think that’s also a useful reminder for developers that ad monetization strategy should be a function of the game’s overall content pipeline, as you said. Because if it’s super hyper casual, people are going to churn out very soon no matter what – they can’t afford to, you know, not be aggressive. 

Jorge: But, there’s one thing there because I’ve seen like some developers that are trying to get into the hyper casual world but they are, let’s say, they love games and they want to produce a good user experience and I say, okay, you cannot do both. So you can do hyper casual and pretend not to show ads because you won’t be profitable ever and ever. So, in order to try to maximise your efforts, you should treat the genre as per the rules. 

So if we take the idle genre, there’s a lot of people that build the game around the concept of rewarded videos, instead of anything else, they think how can I get people to watch more rewarded videos, and they make the game out of these. So I think one example of these is Future Play, games developers from Finland, they say in a couple of press notes and I think in conferences as well, that they build the games around these particular elements – this is rewarded video. 

Shamanth: So they’re like, okay, let’s take rewarded video as a format and build games around that. Interesting, certainly something that could be worth diving into on a different episode. And you know, you briefly touched on this. But how do you recommend thinking about ad monetization strategy? And how it impacts retention and monetization? Like for instance – so if you’re showing interstitials, I imagine users clicking on that are going out of the game – that’s going to hit your retention numbers, but it’s also going to make you money. But if it hits retention numbers, they’re probably not going to buy an IAP. How should a developer think about this trade off?

Jorge: Yeah, so well, there are a couple of ways to think about it because if you have ads within your game, chances are that your users can churn because either they don’t like the fact of seeing ads because it could be intrusive. But they could also churn because somebody has like a cool game or a cool app. So they click on the app, install the game and discover that it is amazing and they like it more so then they don’t play your game anymore. These churn of course, so that usually happens when there’s like a competitor advertising within a concrete game. 

There’s like, kind of two sides of the coin, there are studios that blacklist their competitors right off the bat, like whenever they launch a game, even pre order launch and they say I don’t want these, these, these, these, these, these titles to be shown for my users. Yeah, but I think the correct way to see or to consider these kinds of churn because of competitors – it is looking at “I am making more money because they are advertising or not,” so you need to look at these churn versus revenue ratios. 

And I can have a good example here because in one of the projects I worked on, we saw our eCPM like going crazy high, from 45 to 80, on average, like on a couple days, or I think it was for a period of a week. And then we realised our retention in that country where we had these crazy CPMs was dropping. And then we discovered, it was because a competitor or a game with the same genre as ours, was advertising and was making our users churn through their game. 

But our first thought was like, okay, let’s blacklist these fuckers. But then we said, okay, why should we blacklist them if we are making a lot of money just because of the eCPM that they are giving us because they’re CPI bid is like really, really strong. And at the end of the day, like our users are making them money. So we analyse these ratios, and realise that it was actually profitable for us, let them advertise. And that’s something like developers should look into. It’s not an easy thing to look at but there are some vendors in the industry that can actually provide these kinds of insights. So it is good to take a look at it if you are making a lot of money based on ad revenue. So you need to see okay, it is profitable for me to let these users go, in exchange for a high CPM. 

And I think, for me, that I’m a marketer, business owner, so it is like, I don’t care. I want to make more money. I don’t care if my competitors actually advertise on my game, they’re making me more money. Yes. Okay. Take my users. At the end of the day is like trading so you need to trade them like in the most expensive way you can. So if it’s profitable for you like why not? 

And then the second basis of these retention versus ad thing,

it is trying to analyse what is the impact of the intrusive ad formats like interstitials and banners – do they hurt retention or not? And what are they producing financially in exchange? So let’s say if you lost like 3% in d1 retention because you’re showing a lot of interstitials or your interstitial frequency is really high. But in exchange, your ARPDAU is 30% higher, you’re like, why wouldn’t you be even more aggressive or as aggressive but considering that you need to find the balance between losing retention and making more money. 

For me personally, more money weighs more. There’s people that say like, okay, I don’t want to produce a bad experience. But again, like if we are here to make money, we should take the money. On the other hand, they could also produce like bad reviews. So, most of the hyper casual bad reviews say, there’s a lot of ads, there’s a lot of ads, spammy, spammy spam bla bla bla. So, if your rating overall goes below 4, then it can have effects on the ASO as well. So at the end of the day, it’s a complex thing, but I think the right way to approach it is to test it. So you should be able to test any kind of variable that is related to making more or less money.

Shamanth: Indeed, indeed. As you said, it’s important to keep your eye on the big picture, not just say, “Oh, this is my competitor, not doing it,” but it’s like, what is your ultimate goal? And I would 100% agree that there’s always going to be a trade off between retention and monetization. And, you know, there’s no one single answer, but you need to figure out what’s going to what’s going to meet your eventual goals. 

Jorge, this has been incredibly instructive. I’ve learned a ton – thank you for going into everything in ad monetization in so much depth and detail. And thank you for being a guest on the Mobile User Acquisition Show.

Jorge: Thank you very much for inviting me. It’s been my pleasure and I hope somebody can get a few things out of these and actually improve their businesses. So that’d be awesome. Congratulations for this show. It is always very instructive with knowledgeable people. So I’m very proud to be part of it. Thank you very much again for inviting me. Talk to you soon.

Shamanth: Wonderful.

A REQUEST BEFORE YOU GO

I have a very important favor to ask, which as those of you who know me know I don’t do often. If you get any pleasure or inspiration from this episode, could you PLEASE leave a review on your favorite podcasting platform – be it iTunes, Overcast, Spotify or wherever you get your podcast fix. This podcast is very much a labor of love – and each episode takes many many hours to put together. When you write a review, it will not only be a great deal of encouragement to us, but it will also support getting the word out about the Mobile User Acquisition Show.

Constructive criticism and suggestions for improvement are welcome, whether on podcasting platforms – or by email to shamanth at rocketshiphq.com. We read all reviews & I want to make this podcast better.

Thank you – and I look forward to seeing you with the next episode!