ABOUT DAVID: LinkedIn  | DataSeat




ABOUT ROCKETSHIP HQ: Website | LinkedIn  | Twitter | YouTube

KEY HIGHLIGHTS

⏳ Cross-promos have been around for over a decade

🏡 Why is it becoming important for gaming companies to move cross-promos in-house

💪 How the post-IDFA world has been a catalyst for cross-promos

⇩ When it makes sense to show cross-promo ads even to your high value users.

➰ Why cross-promos within a gaming company should be reciprocal

🔎 Why companies should focus on portfolio LTV rather than gaming LTV

👍 How can first party data help in cross-promos

💭 Why game companies need to rethink their marketing strategy around cross-promos

👀  It’s going to be important to have a consistent, universally understood index for first party data

💣 Why your mediation company’s lack of support for all SDK can harm cross-promos

🔌 What exactly is an ‘adapter’, and how does it work in mediation

📈 The return on opportunity cost as an important metric

KEY QUOTES

How do you spot an opportunity for cross-promo?

They are a highly valuable user, but perhaps they’re on level 40 or 50. Or perhaps they’ve been playing the game for 28 days, and the average lifetime is 35 days. So when you start looking at high likelihood to churn, or you start looking at churn bell curves, there is an obvious opportunity to actually start doing cross-promos to even your own high-value users.

Understanding why a cross-promo didn’t work

The next question I’d be asking is whether you included any first party data in that cross-promotion. After all, the gaming company doing the cross-promo can start including their own data in the bidding logic to decide who to serve adverts to. For example, how long have certain users been playing the game? Are they an ad whale? Are they a depositor? Did they come through a paid or organic channel? Just knowing whether a user typically has clicked and downloaded other games in the past is a very strong signal as to whether it’d be worth serving them an advert.

The infrastructure required to drive cross-promotion

Let’s assume you have SDKs, and there’s ironSource, MAX, Fyber, Admob. You then need a bidder that you actually control that can then respond to those impressions coming from the SDK. And when I say respond, I mean deciding how much to bid – is it worth $15 or $8? So you need a bit of control over that.

How do you define Mediation

So the developer or the publisher needs to be able to operate with multiple ad-serving SDKs. You know, they need to work with ironSource, AppLovin, Facebook DataSeat, Vungle. Mediation is a software layer that manages that process, basically which of the ad network SDKs is called.

The key function of adapters

Any SDK or ad network you’re working with needs to have support from the mediation company. And that actual software support is called an ‘adapter’. It’s a piece of software that goes around the data SDK and allows it to operate in a MoPub environment or allows it to operate in an AdMob mediated environment and so on.

What’s opportunity cost

So every bid we make on cross-promo comes with a price. We still bid $10, for example. We still recognize that we spent, say, $12,000 this month on cross-promo, because by doing so, you can then calculate return. Now it’s not return on ad spend as such. It’s not ROAS. It’s more return on opportunity cost

The importance of calculating return on opportunity cost

One of the challenges of the old way of doing it, which is when we didn’t really record the cost, was that the monetization manager had an issue! He could have got $10,000 from an ad network, and instead they’ve just served a load of house ads and he’s missed his target! So internal financial recognition is important, and the monetization manager knows that it actually wasn’t $10,000 taken off his table. It was just replaced by internal money, which is still calculated ROAS behind it.

FULL TRANSCRIPT BELOW

Shamanth 

I’m excited to welcome back David Phillippson to the Mobile User Acquisition Show. David, welcome back to the show.

David Philippson

Thank you, Shamanth. It’s good to be here. Thanks for inviting me back.

Shamanth

Certainly, very thrilled to have you. The couple of times we’ve spoken, I’ve always learned something new about things that I thought were either obvious or arcane. So I think you have the gift of being able to explain the obvious and the arcane in ways that I think few people can.

David

Thank you. I’m glad your audience has found that useful in the past. I’ll try and do my very best on new subjects today.

Shamanth

Today we’ll be talking about the state of the union, if you will, of cross-promos after ATT, and how the cross-promo landscape has been impacted by ATT. So I would love for you to set the lay of the land and tell us why at a high level, these have become much more critical today for game studios or developers.

David

Sure. I think that is a good starting point because the emphasis on cross-promo has changed. But if I may Shamanth, I’d like to address a couple of other topics before I answer the most immediate question, because I think it is worth defining what we mean by cross-promo within the framework of our conversation.

Cross-promo is not new. Let’s take EA as an example. EA isn’t a customer of mine, but it’s a good example because everyone knows them. So, has an ad network over the last five years been serving EA ads in other EA games? Of course they have, and they should have too, and the performance was good. But that isn’t what we’re talking about. That’s just been good practice over the last 5+ years. That buy would often have been on a CPI arbitrage, and the actual advertiser has no idea of how much money that ad network is really making.

So what we are talking about in the framework of this conversation is taking that capability in-house, where a gaming company runs those activities themselves. This would involve actually blacklisting their own publisher sites with these ad networks and stop serving their own ads inside their own inventory. Basically, they now want to do that themselves, in-house. So that’s the framework of this discussion. Now, to answer your more specific question, why has this become more important?

It’s quite interesting to look at the history of cross-promo. You’ve been doing this as long as I have, Shamanth, so you’ll remember this. Even 10 years ago, Chartboost was heavily promoting and recommending cross-promo and direct deals. But it didn’t survive and thrive.

What happened since then was that UDID became IDFA, Facebook mobile launched, the whole start of behavioral-based targeting and profile-based targeting superseded any performance that would come from cross-promo. So essentially, since Facebook knew everything about everyone, the performance was fantastic. Ad networks were able to achieve the same thing by getting device IDs from MMPs.

Now, cross-promo has become important again because the data advantage that the ad networks had to drive superior performance has been taken away. It’s really become a leveller. So essentially, UA has become a lot harder outside of the walls of a big gaming company. Companies can actually do it inside their own walls and do it better. So ATT has been a catalyst for cross-promo because it has removed data from the third party, whilst the first party still retains some data: some, not all.

Shamanth

That’s a useful way to look at the third party-first party distinction. The phrases ‘third party’ and ‘first party’ have been thrown around a lot, but I don’t think they’ve been explained quite so clearly. And it certainly makes sense that the big ad networks, particularly Facebook and Google, have had so many device IDs that they could easily out-compete any cross-promo you would run, which was certainly my experience as well.

Now let’s assume you’re still running cross-promos post-ATT. My own experience from cross-promos—albeit from pre-ATT days—was that it was just hard to make them work for a couple of reasons. One was that often you just wouldn’t have a lot of affinity between game genres. And secondly, if you were to talk to the game teams, they would be reluctant to show ads to their most valuable users because they don’t want them to churn out. So for those reasons, my personal experience was that cross-promos were just a way to get strong ROAS or see scale.

Given all of this, do any of those dynamics go away post-ATT? And if not, is there some level of scale or heterogeneity, or maybe some sort of companies for which it makes sense?

David

First of all, they’re both very real challenges and objections. They’ve not disappeared. But they are very different. So if I may, I’ll share my views separately on both.

First, let’s take a top tier gaming company, where one studio challenges and says it doesn’t want to serve ads to its depositors. That’s a valid challenge and it does happen. My view on it is this: I would look at that opportunity and ask the studio, ‘Will you serve any adverts to your high value users?’ If the answer is no, then the challenge is actually legitimate. They have a small percentage of depositors, they don’t want to serve them an external advert, and so why would they serve an internal advert just for another studio? So in some instances, that is the right answer. Don’t serve them the ad, since you want to retain them.

But now there’s another variable and you’ve just mentioned it.

They are a highly valuable user, but perhaps they’re on level 40 or 50. Or perhaps they’ve been playing the game for 28 days, and the average lifetime is 35 days. So when you start looking at high likelihood to churn, or you start looking at churn bell curves, there is an obvious opportunity to actually start doing cross-promos to even your own high-value users.

So that was the first thing: If you don’t serve anyone ads, the same logic should apply to these users, but remember to consider churn and likelihood to churn.

The other thing to bear in mind is that cross-promos should be reciprocal. Doing cross-promos is almost like building an ad network within a gaming company. So if Studio A is sending users to Studio B, then Studios B, C and D should also be doing the same, you know? There should be this virtual circle of high value users remaining within the walls. That should be one of the top objectives of the senior management of these large gaming companies. This is becoming more important post-ATT: How do we retain all of our top quality high monetizing users within our walls, within our portfolio of 10 or 20 games?

So yes, it is happening. A lot of these top tier gaming companies aren’t thinking in terms of siloed gaming studios anymore. Or rather, I should say the ones that aren’t thinking this way are going to be more successful. They should just try and retain their users, move them from one game to the other, and make the portfolio LTV as high as possible, as opposed to a single game’s LTV.

Shamanth

That makes a lot of sense. But I think considering the portfolio LTV as compared to the gaming LTV definitely requires significant retooling. Of course, the tools are available, but I just don’t think most game companies think like that.

David

They don’t! Not only that, if you look at their growth and their strategy for growth, it’s often been through M&A. And if you’ve acquired a studio and they’re on earnout now, and they’ve got to hit certain revenue goals, they’re not going to start serving ads elsewhere. So yes, there has to be a strategic rethink on how one game’s contributions to the rest of the portfolio can increase profitability. It’s beginning to happen, and I am starting to see people thinking that way. But the challenges are real and I can’t diminish them.

Shamanth

And what about the affinity between games? How much is that a challenge? I’ve certainly seen many games that on the surface appear to be high affinity games and don’t turn out to be so.

David

Let’s take the extremes. If you had a zombie shooter that had a male skew, and then you had a Match 3 game that had a female skew, that obviously won’t be a good fit on the surface. And both you and I would probably advise them against cross-promos in those two games. But if you had two hyper casual shooters, and they were successful in similar demographics and similar markets, it’s very likely that they would cross-promote well.

But now, if you had that headline experience and it didn’t work,

The next question I’d be asking is whether you included any first party data in that cross-promotion. After all, the gaming company doing the cross-promo can start including their own data in the bidding logic to decide who to serve adverts to. For example, how long have certain users been playing the game? Are they an ad whale? Are they a depositor? Did they come through a paid or organic channel? Just knowing whether a user typically has clicked and downloaded other games in the past is a very strong signal as to whether it’d be worth serving them an advert.

So again, the challenge is real. But once you’ve qualified it down to a sensible approach, where there is some affinity between Game A and Game B, layering in first party data in my experience can then deliver performance and scale, depending on the MAU and DAU of the publisher app, obviously.

Shamanth

Yes, those data points you mentioned—whether they came from paid or organic, are they likely to churn, what level they are at—a lot of that data is definitely available within gaming companies. But I don’t think that pre-ATT, the marketing teams that were doing cross-promos were looking at any of this for segmentation.

So again, what you’re saying is that it really requires a rethink of the marketing strategy or even the company’s overall strategy to make the most of the first party data that could be very valuable in maximizing the value of prospects.

David

I do think so, but the same kind of rethink of strategy and data doesn’t just benefit cross-promo; it can also benefit the organization and monetization as a whole.

We’ve spoken of this in our earlier conversations: In the past, the intelligence was on the demand side, but in the future, the intelligence is going to be on the supply side. Cross-promo is just one example of this, but I can also see a future when monetization managers will start requesting that their first party data gets included in the bidstream. So there definitely needs to be more focus on collecting first party data. It’s obvious why this focus was missing in the past, because you got performance pretty much for free from Facebook collect for sure.

Shamanth

And when you say, ‘Include first party data in the bidstream,’ are you talking about the bidstream for cross-promo or UA?

David

In what I just said, I was referring to both. The benefits of sharing that data with the impression helps cross-promo. But equally, if you’re doing a unified auction out to, say 10 different bidders, I see a future where they’re also using their first party data. Some systems have started coming to the market already.

Shamanth

So if I understand you correctly, on a tactical level, if I were bidding on an exchange for my game, I would tell the exchange something like, ‘I would like women, age 35+.’ Is that what you’re envisioning?

David

Absolutely. Of course, age and gender are just examples, although they’re easier for people to understand. And in fact, this is already happening with dating apps, which are some of the few apps that did have both age and gender in an accurate and consistent word scale. It certainly worked.

But age and gender aren’t always the easiest things to capture from your user base without scaring them away. But it is possible for gaming companies to start looking at what it is that they do know about their users—for instance, are they hyper casual users? Are they ad whales? Do they come from paid versus organic channels?

Shamanth

If you knew those data points, how would you use them for UA? For instance, as you said, a hyper casual user with the propensity to churn in the next seven days?

David

Well, I do run UA for a bunch of other large gaming companies, and if that consistently came into the bidstream, it would definitely become a variable in our own machine learning bidding. We would detect differences between someone who had been playing a game just for one day, compared to somebody who had been playing a game for 60 days. Of course, it’d be different for every game, and that’s where machine learning is a lot better at it than humans, because every one of these games where you’re seeing that data is different. But it does predict the likelihood that a user is going to click and install another game, and there are some first party variables that help that prediction.

Shamanth

So it may be a challenge to find standard metrics that are also first party data that go into the bidstream. Of course, if there was a standardized set of metrics, then it would be easier.

David

I think this is going to be a big topic in 2022. I’m going a bit off my script of cross-promo, but I refer to it as a taxonomy of first party data, because that is what is required – a taxonomy or a consistent index or dictionary of first party data that is shared and needs to be understood by all. And I’m really expecting this to emerge not just for cross-promo but for all marketing activities.

Shamanth

I’m just trying to wrap my head around this idea – if there’s a consistent taxonomy, does that in some way become third party data? Let’s say I’m a DSP. I know there are 10 users who have a high propensity to churn in Game A. I’m then making use of that data to place bids for Game B. Does that not automatically become third party data?

David

I like the way you’re thinking!

If that company or ad network or DSP stores the data and starts building profiles on it, then yes, it starts getting into this gray area of concern. However, in the bidstream where this data can be shared it will be a non-persistent identifier. There is no IDFA there that allows you to build a profile.

So the decision is in real time. And yes, you can imagine that there is a bit of data, and it includes that this is a hyper casual user who has a high propensity to churn. But I can’t build a profile on it. It would just increase my bid price or increase my prediction.

Shamanth

Sure, but there’s nothing to stop them from building a profile, though.

David

Well, you do need a persistent identifier to build a profile. If it didn’t have an IDFA, you can’t build a profile on an IDFV. Even if I recorded it, you cannot see the same user on another publisher.

Shamanth

What are some of the pieces of infrastructure that are essential to drive cross-promos that use much more of the first party data? It was always available, of course, but it’s now going to be increasingly used.

David

Absolutely. So first, let’s assume the previous topics are aligned, which means there’s a gaming company with multiple games, and there’s some affinity between those games, and there’s quite high MAU and DAU. All this means it’s worth embarking on this project. The first thing you need is to work with a partner that has iOS and Android SDKs, and those SDKs need to be supported by your mediation stack.

The typical mediation companies – ironSource, MAX, Fyber, AdMob cover 99% of what’s out there with respect to mediation. Now, these mediation companies don’t support all SDKs. My view is that cross-promo ads should be in the monetization stack and should compete against other external revenue. But for that to be the case, it has to be supported by mediation.

So

Let’s assume you have SDKs, and there’s ironSource, MAX, Fyber, Admob. You then need a bidder that you actually control that can then respond to those impressions coming from the SDK. And when I say respond, I mean deciding how much to bid – is it worth $15 or $8? So you need a bit of control over that.

What’s unique to some of the other alternatives in the market, you will see some cross-promo ads being done outside of the monetization stack. I’ve started seeing some thumbnails that are specifically for cross-promo. Personally—and certainly with the clients I work with—I don’t agree with that, because I think that any advert that performs for cross-promo should compete against UA. Because if it performs for cross-promo, it can probably perform for external UI and be making real money.

So if you want it to perform within your monetization stack, within your waterfall or amongst your bidders, you need SDKs that are supported by your mediation company.

Shamanth

And for those folks who may not be aware, how would you define mediation? And another word you used the last time we spoke was ‘adapters’. So tell us what mediation is, what are adapters, and what is their key function in all of this?

David

Let’s imagine an awful world when there’s only one ad network and it’s called Google. In that case, you’d only put one SDK in and you wouldn’t need mediation. But that isn’t the world. There are many ad networks with different demand with different amounts of money to monetize your inventory.

So the developer or the publisher needs to be able to operate with multiple ad-serving SDKs. You know, they need to work with ironSource, AppLovin, Facebook DataSeat, Vungle. Mediation is a software layer that manages that process, basically which of the ad network SDKs is called.

This is done through two methodologies.

The first is a waterfall, which is where an ad network will get an opportunity to bid at $25. Ask Facebook if it wants to bid at $25. No? Right, then go down to Google at $20, and so on. So different ad networks will have different opportunities to bid at different prices. This is referred to as the waterfall, and it’s managed by mediation.

What is replacing the waterfall is what was referred to in the very beginning as ‘header bidding’ in the desktop world. It’s become in-app bidding or just ‘bidding’ now. And this won’t go through a waterfall. There’ll be one auction that goes out to all participants on the demand side. And whoever bids highest wins.

People often have a hybrid of these solutions in place today. And for both of them,

Any SDK or ad network you’re working with needs to have support from the mediation company. And that actual software support is called an ‘adapter’. It’s a piece of software that goes around the data SDK and allows it to operate in a MoPub environment or allows it to operate in an AdMob mediated environment and so on.

That’s my quick take of mediation, although you could run a whole show on it!

Shamanth

So mediation basically lets the ad monetization SDKs plug into it and show ads. How does that work for cross-promos, because you don’t need an ad monetization SDK?

David

The core function of SDKs is to serve an advert, whether it’s a rewarded video, an interstitial or a banner. So it means to send an impression out and get a response if they’ve won to serve that advert. It doesn’t really matter whether the campaign behind that advert is a cross-promo campaign or whether it’s an external campaign from Paddy Power or Zynga. It really doesn’t matter. The function of the SDK is the same, which is to serve an advert.

So actually there is no difference between a cross-promo SDK and an ad network SDK. It just responds with a price and an advert, if it’s won.

Shamanth

But it would need to be configured differently if it’s an internal ad?

David

It’s an interesting topic because we start to look at price. I’ll go back to my EA example (just to be clear, EA is not my client, it’s just a great example). If internally, EA were serving EA ads inside EA games, it would be crazy, because they would just be paying themselves, right? No one would benefit other than the banks!

However, that bid price still has a cost, only it’s an opportunity cost.

So every bid we make on cross-promo comes with a price. We still bid $10, for example. We still recognize that we spent, say, $12,000 this month on cross-promo, because by doing so, you can then calculate return.

Now it’s not return on ad spend as such. It’s not ROAS. It’s more return on opportunity cost (turn that into an acronym if you wish!).

So, our SDK still doesn’t behave any differently. If it’s a cross-promo ad, we would still bid with an advert with a price, and that price is recorded. The only difference is that money won’t be merely moved from the company to itself. But the cost—the total CPM and total media spend—it’s still calculated to calculate return on opportunity cost.

Shamanth

Right. And in my personal experience—which is admittedly not very recent, but certainly not post-ATT—is that I don’t recollect us having calculated return on opportunity costs or treated it as a real cost to measure the efficacy of cross-promos. How widely do you see that kind of accounting practice being followed?

David

I think now there’s more emphasis. A lot of people know it’s necessary at the high level. So they’re then dealing with the challenges.

One of the challenges of the old way of doing it, which is when we didn’t really record the cost, was that the monetization manager had an issue! He could have got $10,000 from an ad network, and instead they’ve just served a load of house ads and he’s missed his target! So internal financial recognition is important, and the monetization manager knows that it actually wasn’t $10,000 taken off his table. It was just replaced by internal money, which is still calculated ROAS behind it.

So to answer your question, is this being done at scale everywhere? No, because this is quite a new topic. But to make this successful across the board in more and more large gaming companies, there definitely needs to be financial analysis that comes from it. It should prove that it is worthwhile, which is why I think it has to compete against other external revenue sources. And it should still come with a CPM and a total media cost.

Shamanth

Certainly, that’s definitely a necessity. It’s a change in how you account for your marketing metrics. It’s an accounting change as much as it’s a process change and a technology change, and I can see why it is critical to make a lot of these changes to really see the potential from cross-promos.

Well, David, I’ve always learned from our conversations and this time’s certainly no different. Thank you for sharing everything that you did today, about this goldmine of first party data that we’ve all been sitting on—it’s just that we didn’t notice it since third party data is so ‘easy’ to get.

I’d say this is a good place for us to wrap up, David, but before we do, do tell our audience how they can find out more about you and everything you do.

David

You can find me at david@dataseat.com on email or on LinkedIn as David Philippson. And it was a pleasure to speak about this. So thank you for inviting me on your show again, Shamanth.

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, Google Podcasts 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!