Real-world data and the implications of IDFA deprecation – from GameMakers UA Coffee Talks

We talk about real-world data and the implications of IDFA deprecation.
There has been a lot of VERY BAD information being published on this topic. So we have discussed some real data about what’s actually happening.

Hello everyone and welcome to ua coffee talk. Today we’ll be talking about real world data and implications of idfa deprecation. Now there has been a lot of, and if i’m being really honest, kind of bad information being published on this topic. So today we have with us three real world practitioners and some real data points about what’s actually happening and more specifically the discussion today will focus on:

  • real world data and impact of idfa deprecation
  • specific changes to how we do ua and marketing based on the effects of idfa deprecation
  • implications to product and monetization design
  • and finally who are likely winners and losers from these changes 

And with us today we have first Brian Bowman, CEO of Consumer Acquisition, who has managed over 3 billion in creative and social ad spend from the world’s largest mobile apps and performance advertisers. They see about 150 million dollars a month if i’m correct Brian. Running through their admirals platform, and work for companies like Disney, Superhuman, Rovio, Roblox, Glue,  Mobile jam city and many others. I believe you guys have done some stuff with applovin as well right?

Second is Matej Lančarič, an independent UA and marketing consultant.  Matej is a user acquisition expert with eight plus years of experience specifically in gaming. He’s also globally launched 28 games and oversaw over 20 million dollars in profitable spend across different genres of games.

And finally we have Warren Woodward, co-founder and chief growth officer at Uptick. Warren formally led multiple gaming UA divisions including for companies like Nexon and Wargaming his company uptech offers a full stack growth marketing services and marketing automation tools platform that powers multi-million dollar growth portfolios for companies including Candy, Writer, Gsn , Experian and Mino games.

Welcome guys and I am so glad to have real world experts who are running stuff in actual practice with us today. So I thought we could go ahead and start with the first question which is really around what does the data tell us about the real world impact of idfa deprecation and how does that fit against sort of theoretical implications?  Now I know Brian you have published a blog post showing some game publish were down by let’s say 15 or so to game revenue but do you have any updates in terms of real world data what are you seeing right now?

Yeah good to talk to you today. You know there’s a bit of what we have and then there’s a bit of what’s going on at the public company. So I think both of those are interesting data points. The kind of sledgehammer which we’re seeing revenue down for ios about 15 to 20 % across the portfolio in companies that we’re touching. So it’s a pretty significant impact but when you peel that a little bit you realize there’s a nuance to the way that actually is rolling out where there are different impacts for sub genres. Those games specifically are apps that target narrow audiences they’re hunting whales or those that are really looking for in-app ads. That’s the way they monetize,  those are the most impacted and one of the things I found fascinating from the series of companies that have reported earnings most recently is they’re kind of distilling the impact into two things. One the uncertainty around COVID 19 and the fact that people are going back to work. You know things are getting back to normal, is a way of saying that year-over-year traffic is declining. On top of that apple’s idfa changes and the potential for continued wide range of outcomes both positive and negative so it’s really fascinating when you kind of peel through some of these earning reports like ironsource says. I agree with this the industry thought that there would be impacts in q1 didn’t happen, then q2 it really kind of happened but a little bit only at the end and they’re continuing to monitor for short-term effects. And Adam who’s the CEO of app11 said they see about 80 percent of ios devices are now updated to 14.5 or later. The consent rates are different obviously app by app somewhere between 25, 20 to 25 on the low end 60 to 65 on the high end, which means about 35 to 40 percent of people are opting in to share information so I don’t know if that’s above or below what expectations were but certainly it’s having a material impact as we talk through the rest of the day. And then zanga I thought was a bit more transparent in what they’re seeing so they’re saying that idfa is having a material impact on the choppiness of its business they warn that second half bookings could be down 100 million dollars. They’ve scaled back UA on ios and they’re adjusting ltb calculations to compensate. So lots of really good data when you read through the earnings. When you look at facebook they’re saying something I found fascinating which is impressions are up six percent CPMs are up 47 which is a fancy way of saying that click to install quality is down you know.

Let me pause there and hand it to other folks and see if they have something. That I think that’s a great way to set the table Brian.

I think there’s one important earnings report that we didn’t mention yet, which is apple’s own earnings report. So one thing that was interesting is they are projecting their own services revenue is going to be down for the second half of the year and while they didn’t spell out specifics I definitely know from some sources I have inside there, that they have finally started to realize what some of this data means for their future business impact. I don’t think there is a plan yet as far as how they’re going to react to this but I think it is a relatively recent acknowledgement that Apple is starting to see that some of this data will have a material impact on their own earnings and broadcast out to the market also.

And also yeah I talked some publishers and developers and I think they saw even bigger jobs in terms of revenue on ios since the start of or in the end of the April. But smaller devs and mid-sized ones they got a bit scared and moved their budgets to android already. So you know until they figure out how to evaluate ios properly. They got from spending some amounts of money on ios to spend the minimum until they figure it out so you know trying to find out which schema works and which doesn’t and stuff like that.

Yeah that’s an interesting point and I know Joseph one thing you wanted to talk about here is sort of like theory versus practice. And one trend I’ve noticed in kind of the whole lead up to this and you know after this the switch flipped is there was a lot of theory talk. There was a lot of people trying to know share that they had the optimal model or ways to work within this infrastructure. But at the end of the day the most common thing when I’m talking to to other teams. But what are you actually doing usually it’s more often than not they’ve just paused spend or brought it down massively with just sort of enough inflow of data to to mess with it and to tune it. And then I guess lastly I also wanted to echo Brian’s statement we’re definitely seeing in our portfolio a general rule of thumb is the more broadly appealing. It is still able to maintain some sort of reliable performance in in this environment the more the more niche it is the more difficult it is it’d be a very tough time to be like a an mz type of product right now.

 Yes or in before the idfa, there was probably every week a podcast with some folks, where everybody was saying „yeah we shall see, we don’t know yet, but this is what might happen…“ Come on every week, almost every week!

Right, yeah and in terms of that aspect and Brian I know you mentioned that certain sub-genres of games are being impacted differently and I think the theory has been that the games that are focused more on well targeting. So for example casino games. That would have a bigger impact to those sub-genres and without naming any specific companies do you guys have any data points that you guys can speak about? By the way from my perspective that’s what I’ve been hearing as well or seeing is that to the point about more generally appealing games. Some of those games have actually seen increases rather than decreases but what we would have thought conceptually in terms of the well-targeted types of games seem to at least from a few data points that that I’ve heard about have had a material impact to them.

So a couple things right. Not just trusting what we’re saying. Please you know do your own research. Matej came out with an interesting part which is – they’re seeing social casinos and strategy games being hit the hardest by idfa loss to kind of layer on that we think hardcore and hyper casual are also being negatively impacted. When you look at what advertisers are doing it’s a natural reaction they’re going to try and move spend as much as they can to any viable network which is driving up CPMs on those networks. So those guys that thrive and survive on low-cost CPMs are finding it very difficult to compete regardless of whether they’re a game or an app. We have other companies in our portfolio and they also are finding the impact from CPMs. So I I think it definitely is is having a negative impact on the industry.  The question is you know what do you do to get out of it and what are the next steps.

Right and just to support this, the app annie report I actually just did a quick data look up last night and I kind of published it on my game makers newsletter where looking at social casino and specifically what we saw kind of through may, june and july is that revenue was relatively stable but we definitely saw a decrease in terms of downloads. When you look  at the portfolio of the top social casino companies in their number one game and kind of aggregated that up it does seem there is an impact at least to downloads. And then we would expect revenue to trail downloads by a few months so it does seem there is an impact there as well.

What we’re seeing happening is people LTV models are built up for a long time horizon. A year or two with forecastable returns. When you shake that up and you can no longer measure things after 48 hours what we’re seeing is people obviously drop spend. But then there’s this weird uptick that’s going on in organic monetization obviously. Idea loss is forcing people into the unknown bucket. So how do you build a model around that level of uncertainty? It’s almost CPG advertising where you’re going back to Nielsen clusters and  just the loss of efficiency is pretty profound as we start to realize what this means. In the sustained long term because I don’t think we’re at full capacity yet and I think not until the end of august will things be around the 80 percent. So I think it will continue to get worse over the course of august before it will possibly start to get better.

Right and in terms of the theory there was some thinking and I’ve speculated about this as well that the games that monetize more broadly, and I know Warren you made a similar point, should potentially or could potentially win in the new environment. So you know when we look at a supercell games like  Clash Royale ,Clash of clans, we’ve seen a pretty dramatic increase and they’ve seen the biggest spike in over three years. So from your perspective when we see something like that do we think we can attribute that kind of a change to idfa deprecation? What do you guys think in terms of other potential winners from the new environment and whether we can make any conclusions about whether these kinds of spikes are due to idfa deprecation or not.

Yes so I think there’s a couple of different intertwined things going on here and part of it ties into what Brian just touched on. So one thing I think that powers the trend you just referenced Joseph is if people want a game they’re not gonna  spend again not play a game because they haven’t seen an ad lately so what does that mean is the the games that have more presence of mind or maybe that people played more in the past I think those brands have more power than normal. Another factor there is I know sometimes some of those bigger brand names have a tendency to buy on more blended models, all in models and if that’s the main KPIs you’re looking at a lot of smaller to medium -sized studios have more of a laser focus on more the UA direct performance and those numbers are going to look worse because as Brian alluded to, there’s going to be a higher percentage in IOs that’s not tracked in the same capacity. So I think those are two factors that are they’re leading it you know just the the presence of mind when you know when there’s less effective marketing and then also you know that buying on those blended models. One thing I really want to emphasize that Brian touched on is the model itself that’s a really big area of work for us right now with our portfolio of developers it’s always an area for work with us and part of our philosophy is building a model that factors in correlated uplift. Basically the halo effect of the paid marketing efforts not a blended model where you claim everything but just correctly correlating the organic impact of of your paid efforts and we had something that worked really good before this change happened, but we’ve had to throw that out the door and rebuild from scratch with this. And that’s one takeaway – I would encourage all people to think about whatever model you were using for IOs before june that model is no longer the right model and if you’re still trying to hit those targets, if you’re still factoring organics the same way it’s it’s not going to work. So everyone needs to be putting a lot of focus into the data science side into building a new predictive model for that given product and it sucks because we don’t have very much data to feed it for it for any of us right now. But that’s something that we should all be mindful of and I think part of this is we just as an industry have to buy a lot of data right now and rebuild those models.

Yeah well regarding the Supercell and Clash Royale and Clash of Clans spikes – I believe it’s connected to the idea phase and  as we mentioned the spent went to android. And  honestly  IOs looks the whole ui space last year when COVID started a  very low competition. For example I’m seeing for mid core game  well CPI is not  a pretty  interesting KPI but still for us for midcore game seeing  four dollar CPIs versus  12 or 15 on Android. That’s super big difference and the bigger companies try to to push IOs 14 campaigns  on as they can just to get as much data as they could. They could build those LTVs models and as you mentioned why the well oriented games are suffering it’s because now you need to look at the day zero even first 24 hours and then based on those two first 24 hours build these type of LTV predictions and LTV models which is definitely super hard for the the weather games.

Right and Matej I think that’s an interesting point that you were raising about  the smaller to mid-sized developers might just be pausing because they’re trying to get a feel for the market and  maybe as they’re pausing, companies like Supercell might have an additional effect where because of the lack of competition Supercell’s able to gain more share.

And Brian one of the things that you mentioned in your last blog post had to do with kind of a lot of big changes where you were seeing to the algorithms themselves deployed by Facebook, Google, Tik-Tok causing a lot of instability and dramatic changes in CPM. Do you have any updates on that front is that changing and will that cause a lot of strategies to have to continue to shift out as the algorithms continue to shift as well?

Yes Facebook yesterday put out a an interesting article that I would encourage the folks listening to read it’s called „Privacy enhancing technologies and building for the future“. We’re seeing CPMs are very out of what I think it’s a natural reaction where people are bailing on IOs because of the lack of optics moving into Android which forces up costs. But at the end of the day really if you just take as a statement of truth there’s more impressions in these marketplaces. There’s more traffic being issued in the marketplaces and click to install is down that means traffic quality is down. And what if you were a network what would you do. You would evaluate and optimize your algorithm that it maximizes your own internal CPM . Logical reaction to loss of efficiency that means on the consumption side it gets more expensive for us to buy those people which is why the folks on the extremes are struggling with it. I don’t see that changing in the near future. I think that’s a long-term change that is going to happen how we do next in this privacy forward world. Is it contextual advertising, is it me of the things that were laid out in that article. I mean these are 18-month initiatives they’re not Q4, black friday things are magically gonna change if you’re a public company you cannot spend ,  you’ve promised a certain growth of metrics to hit the street. You’re going to spend through it, you’re going to do it inefficiently and you’re going to try and amortize those costs. Is it happening? Yes. Will it change in the short term? No. Is it material? Yes.

And if you look at the byproduct of what was mentioned earlier with the loss of idfa you’re seeing lookalike audiences start to collapse whether you’re talking specifically facebook other algorithms that are using app events with a limited amount of people opting in to be tracked. That means the custom audiences are very small where an audience used to last two or three months and we get a lot of efficiency out of it. It’s now two or three weeks the even the pockets of efficiency that we used to use to drive long-term kind of sustained success are fleeting. They’re harder to find and they’re shorter. I think that’s a long-term challenge in the way that we’ve all grown accustomed to this kind of laser efficiency with deterministic tracking

One quick comment to that – of course the facebook performance is decreasing but quite heavily I think. Since the  beginning of march even, and it’s getting worse and worse in the last couple of weeks and months. But I’m not seeing the same thing as you mentioned with the look-alike audiences. It is actually hard to build those of course and you need to refresh them pretty often but it’s still on. These audiences are actually working better than the aaa campaigns and other types of targetings which is definitely surprising for me.

Can I al quickly get everyone’s opinions on what you’re seeing on google? I think a lot of this discussion so far has been focused on facebook but any thoughts there as far as how you’ve seen them adapt?

Well for me google UAC is performing better than facebook and I’m trying to move there many of IOs, not much from Android. But yeah the IOs is struggling a lot.

I was gonna say for us it’s been it’s been horrible. I’d say  google has of the big players, they’ve done the worst of adapting. It’s this weird system where they report on the google modeled installs and we’ve had all these conversations with them where they’re like „what are you talking about it looks great to us with the  the google installs“ and we are like „yeah but that’s just your own black box logic“ and when you  use a neutral judgment system of an mmp it looks completely different. Until google gets their act together and provides me with better modeling, we’ve pulled a lot of that uac IOs spend.

That’s true and they were late with the sk network integration and  all you see, the conversions on IOs campaigns  with five day delay. Seriously? What can you do in five days? just it’s terrible,

We’re seeing things a bit differently  just to hammer home the look-alike audience thing. We’re talking a broad swath of clients here that are kind of spanning the genre segments here and entertainment and e-commerce  it’s not just gaming related. It is a material decrement in performance,  half of what we’ve seen pre may 26th. Google in my understanding of the algorithm is they were probabilistic in their nature, it’s less impactful but we are seeing a decrement to it. The one kind of bright spot is Tik-Tok that seems to be doing better. It could just be that many advertisers weren’t in the platform and now they’re  starting to move into it but google is still doing better proportionally for our portfolio than other platforms. We’re able to push into it and we’re able to sustain now, to be clear it’s not as efficient as it was but it’s not as negative as other platforms

Well for me the clear winner of  the last couple of months is Unity. It’s just that I’m  surprised how Unity was able to improve the performance and prepare for the idea phase and everything. It’s just really amazing how things are going there. As well if you read the Google earnings and you just do a quick search of it. Idfa didn’t come up one time. That was just, for whatever it’s worth, that was interesting and shocking to me. 

I think it both Iron Source and Unity use this phrase just to toss it out to folks. Contextual advertising whatever that means to all of us seems to be the direction in which the world is headed they both feel they’re well positioned to take advantage of that. It was just absolutely staggering when you read the footprint of what these guys have. In fact I think Unity raised their 2021 guidance by 50 million dollars. They’re bullish on where they’re headed and the stability of their network and their lution with idfa.

That’s one of the things we personally haven’t done by the way. Unity captures and analyzes 50 billion in-app events each day. That’s just I mean it’s mind-boggling the amount of information we have. Because I think at the end of the day we’re all going to have to go wherever there’s pockets of efficiency and move money around as aggressively as we can.  I’ve slightly converted conversation half a year ago or last year I would say. Oh well facebook and google well that’s enough to manage half a million millions of spend. Now it’s okay not anymore you need to add many sources to get efficiency for different perspectives.

I was going to say I have a slightly different take here. We do a lot on the SDK networks  and we’ve definitely seen on IOs for certain apps that the sdk networks in this era have started to represent a bigger portion of the profitable spend than facebook, than google on IOs right now. 

But I’ve seen a different trend and reason we said we won’t run campaigns on sky network yet you have to enable probabilistic through your mmp to run and I see that’s what’s powering a lot of that ecosystem still. 

It’s working now but I think that’s gonna be at-risk. Kind of short-sighted approach depending on how apple views the spirit of that and maybe as a high level just to kind of characterize the row as ROI loss from idfa deprecation bit more.

Do you guys have any high level thoughts or kind of benchmarks in terms of what percentage loss that you guys are seeing across the different paid social channels? Like: that’s my generic 15 to 20 right 15 to 20?

If you push back on it and you look holistically that’s the number if you then you zoom into a genre it all changes right because it’s nuanced based on how are they whale hunting. Are they looking for high-end subscribers right all of those businesses should be negatively impacted from the reduction in people opting in to be tracked that’s a natural byproduct. 

Those folks  you mentioned that are kind of match three or very generic are impacted but less  because they’re spending and they pick up the loss in organic and it’s attribution analysis. We can figure it out, it seems to be different if you’re overall maintaining the same pool of people you just can’t attribute them necessarily either to a network or to a source and obviously IOs 15 deals with me of that at the network level but inside of the look audiences it’s unrecoverable. Matej warned you guys!

If you push back on it and you look holistically that’s the number if you then you zoom into a genre it all changes right because it’s nuanced based on how are they whale hunting. Are they looking for high-end subscribers right all of those businesses should be negatively impacted from the reduction in people opting in to be tracked that’s a natural byproduct. 

Those folks  you mentioned that are kind of match three or very generic are impacted but less  because they’re spending and they pick up the loss in organic and it’s attribution analysis. We can figure it out, it seems to be different if you’re overall maintaining the same pool of people you just can’t attribute them necessarily either to a network or to a source and obviously IOs 15 deals with me of that at the network level but inside of the look audiences it’s unrecoverable. Matej warned you guys!

I have a question – do you think what kind of laws would we see if the spend stay the same because the spend was really low at in the last months?

Yeah but just forget about spend for a second again you’re a network right you’re in charge you’re the CRO of a network and you’ve lost your ability to deliver people at a specifically high revenue. You’re going to put more low quality traffic into the mix to offset the ability to target on the high end. It’s a natural way of doing it to maintain your margins. I think just the ecosystems reaction to the loss of specificity is going to be you’re going to have to buy it you’re going to have to spend through it you can’t target it. There’s an advantage as a network which is I can maintain my own gross margins because you’re not going to be able to find more efficient traffic in another source that I would bet on.

Just to add my take here. I think that the general numbers that Brian provided are roughly what we’re seeing across the portfolio and we bias towards gaming we definitely have me non-gaming in our portfolio as well the one area I would diverge a little bit is again part of our work is building the right predictive buying model for each of our developers and we’re actually seeing on apps if we haven’t essentially got buy-in to a new model and we’ll come to the developer and say hey with the trends we’re seeing we think here’s how we should adjust the targets based on how we see opt-in opt-out rates based on how we see  the correlation of organic revenue,  if we’re still working off the old model basically if we haven’t bought in mething new we’re seeing organics trend down too because we know that good UA has a halo effect.

And  if you’re taking a big step down because you’re only going to those pockets where you can still hit last year’s goals then you’re losing that halo effect too soon. Some of our really large scale titles have had actually a negative impact to organic for most of them for small to medi size. It’s it’s been the opposite. The organic represents a higher percentage as there’s more there’s more loss but on bigger titles we’re seeing  negative organic impact.

And maybe now we could kind of shift focus to talk about specific tactics approaches to how you guys as performance marketers need to change the way in which you do UA based on kind of the things that we’ve been talking about and I know Warren you talked to me before about ways in which with even idf deprecation you can try to model campaign performance and  could you speak to what have you been doing to estimate performance or how you’re changing me of the models that you have to try and  operate in this new environment?

People are gonna have to build up I mean we’ve had it easy on the in the in mobile UA for the last the last several years  you could just look at the n bers and know you could basically operate off of them  any team is gonna have to get into the game of prediction now whether you’re comfortable with it or not and one thing I would recommend everyone do today is just one easy actionable if you’re not tracking opt-in and opt-out events yet just work with your product team and get those implemented. You might not know how you’re going to use them yet but in your future models you will be factoring them in some way.  I’m not going to get too into specifics of sort of what the uptick model is but I will say that each product is going to vary depending on how you’re surfacing prompts to the user depending on what your demographics are but you will want to factor in those opt-in opt-out rates and you can use that. We all know that there’s a lot less that’s tracked now right and we want to quantify that and mix it with what we can still track and have confidence in and use that to build a model. And it’s even messier because it’s not a constant. This is flowing over time, we know that adoption rates have changed over time, we know that even the trends of the percentages of opt-in we’ve seen changing over time you have to have dynamic inputs into the ideal model here. I guess what I’m saying is there’s not one size fits all model here but identify what are the variables that you would feed into your model and make sure that you’re starting to implement those events and track those now. You can have your data scientists  three months from now start working with that data and you build and back test several different models.

All right Brian and Matej do you have different ways in which you guys are kind of modeling performance?

We’re working with predictions and that just to simplify it’s using an appropriate multiplier for each cohort and that changes on daily basis anyway. And this helps us to get the sense of what can we expect at least and I mean obviously the whole methodology changed much. But even if you don’t have the prediction models, the fact that a lot of developers decrease the spend on IOs can actually help them  better identify the campaign performance even when you’re looking at the blended drawers, numbers or the blended numbers, since the baseline of the organics dropped when you drop the budget. You can clearly see what’s happening after you just increase the spend as well.

That’s a good point. I just want to say we have an opportunity right now because usually when you’re trying to model uplift you always wish you had a period where spending just off. You see a real baseline. Note if you have a real baseline opportunity and if you’ve turned IOs off use that as an opportunity to baseline out your organic revenue. 

The fact we’re taking I think is a bit different here which is we’re not of the belief that we’re going to be able to look inside of the walled gardens of facebook or google or tick-tock and we’re not worried about modeling in many ways we rely on our companies  our advertisers to do that. The things that are working at least in the con constraints that we operate in on paid social are top country  the things doing the strongest performance top countries localized ad copy with aeo ideally with broad targeting interest groups are definitely making a comeback  those things where you can in essence formulate a contextual mapping  if you push back for just a moment and you think there’s two kind of driving forces that have been impacting the industry for the last couple of years  facebook and google with full media buying automation at least their push into that  uac and aaa.  

If you say it backwards they’re removing levers from media buyers to simplify the process and level the playing field between advertisers big and small but as you eliminate levers you eliminate the things that make it efficient to do. If you then layer on top of that idfa removal  deterministic to probabilistic tracking it’s a one-two punch and  at the end of the day we believe that the only primary lever that we currently have we the industry is creative  in terms of maintaining and driving successful return on  ad spend it’s just really hard to do well right a vast majority of creative fails it’s about 85 to 95 failure rate meaning do you beat the best video or asset in your portfolio odds are no when you do get a winner it lasts about 10 weeks  there’s a brutal cycle of always working really hard to stay on top of the creative fatigue one of the challenges we’ve realized. And if you guys have ever seen the movie moneyball with Brad Pitt –  I love the analogy – which is quantitative analysis overlaid against recruiting for baseball – I see that exact same thing going on but on madison avenue  mobile development teams are very focused on revenue LTV and retention but they’re not focused on creative fatigue and because of that their marketing and product teams are often kept separate they’re not forced to work together the assets the marketing team has is often behind for advertising that is behind the curve. 

If you don’t fundamentally pivot the model and squish the teams together and come up with a new way of working you’re going to be a dinosaur. The net impact is creating the ability to say what I read watch and listen to my marketing segmentation analysis for my user motivations coupling design with onboarding flow and then with  interest groups seems  a sustainable long-term way of doing this it’s just hard. Well it’s definitely not easy but that’s the industry. 

Brian if I’m hearing you correctly it sounds you are primarily focused on top of funnel to install conversion but  are you not doing anything with sk ad network and maybe I could ask the other guys in terms of  sk ad network conversion value data how are you guys leveraging that if at all are you guys  mucking around with that stuff to try and optimize your campaigns well I’m pulling those data to to get them to the prediction models and trying to use it that way basically from the prediction models then optimize campaigns?

Okay so to answer your question no. We obviously always focus on the bottom of a funnel but here’s the problem okay in this new world because I didn’t layer in idfa plus media buying automation plus when you only have one account right the restriction of scan is you can only have one account running IOs 14 6 and later almost always the advertiser is going to internalize that account right there unless we’re the agency of record it’s infrequent that we will get access to that account  that addresses a bit of the problem here  we do younger older rather operating systems on IOs and Android  it’s more of a focus on where performance advertising currently exists and then on the IOs side the only thing we’re going to operate on is because the lever you have is how do you drive I  the phrase earlier  quality UA it’s got it’s creative and it’s creative to something you can map it to for targeting and that’s contextual in some form.

I wouldn’t say  increased performance or increased importance of creative but just more  people people can’t ignore it anymore me teams looked at it as more optional or you could get by we approached it in a different way for our company the first thing we had our engineering team put re urces on was actually building  an a creative tool and people  a lot of people that know me thought that was weird because my background’s always in UA it’s  why are you guys building a thing first but we know that  yeah the levers we have as UA practitioners are getting fewer and fewer and  we wanted to build a really robust system to just  rapidly test a bunch of creative and  auto optimize the visuals of the app store and have that be a living breathing thing that you could improve over time keep throwing more creative at it and  that was the first actually product that we shipped and built  and it was definitely a big gamble but we are seeing it pay off and we’re starting to see  our both the developers we work with now and outside of our circle just starting to be  crap we really don’t have a lot of levers maybe I’ll finally do that a thing that I’ve been putting off that  I know I should do  that and getting people to think that  it’s not just about  running an icon test  you got to treat this the same as you treat  your ads portfolio of  be constantly testing constantly innovating  constantly refreshing  and  that’s definitely been another  another lever of creative that I think people haven’t  put enough importance on and that al benefits your organic conversion too  that’s been another way that we’ve kind of doubled down on on creative.

But the quick question is  do you remember last year we had  the podcast about death of performance marketing and you asked me  okay is it going to be really dead now? Dude we have more work than ever really we need to figure out  many things  many things really  very far from being dead. I always say we’re  the plumbers. You call them we only you hear from people  when the pipes burst when the toilet’s clogged and it’s  we need help right now…

Brian you’ve probably seen similar  but yeah just the demand for help has never been higher but just to dig a little bit more deeply maybe I could ask again in terms of  ski network conversion value but is there anything specifically that you would recommend in terms of trying to do with respect to how you how you treat that and then I’ve al had this hypothesis which I’ve talked about last year where I actually based upon  the kind of game that you have I actually think that doing stuff with sky network conversion value probably benefits  short rpu games meaning  games in which your rp curve kind of flattens out relatively short quickly but for the well-driven the more longer rp type of games it’s  the predictability is just going to be  completely unguessable in my opinion but you guys have any thoughts on that as well.

Yeah I can say for me  maybe it’s a little contrarian but we’re intentionally not investing  we invested enough time to research there to think that we actually don’t think that’s going to be the long-term viability viable lution for for profitable UA  we’re building out a different approach that’s not leaning on that as the main layer I don’t know if it’s going to pay off and  I’m not going to speak to the specifics yet but I think that apple has made such a high barrier of entry and such a usable set of tools that either that’s gotta evolve or we as an industry have to build superior approaches to that yeah we spend enough time on it to say  we don’t think this is gonna be super tenable in the long-term effective path and we put our r d efforts elsewhere okay which is what I was trying to highlight when I say you’ve got to completely change onboarding flows but coupled with that is if you don’t have an early signal that meone exhibits behaviors that they’re going to be a high payer that I mean that’s obviously not an easy challenge for for design  app designers because you have to fundamentally change the dynamics of the game to get meone to kind of push to to to monetize early on yeah that I think is a very difficult challenge right if otherwise if you could you would right if you if you can get people to pay earlier we all would have done it years ago okay of course it’s not that easy to just put some special offers in to day zero and then not up all the economy in the game right.

Yeah and maybe just to push a little bit more Brian on your point about creatives  any thoughts in terms of how creative testing has been impacted on IOs yeah I mean IOs is dead you can’t creatively touch  just don’t waste your time  it’s all it’s all moved to Android  we we’ve published a lot of white papers on our opinions on how creative testing works on the networks there is no button you can hit in any of them that says treat my best ad as if it’s day zero to remove all history from the bias of the algorithm and let’s do a true a b or multivariate split they always drag with them creative history  we it’s a three-step process and  we run people we run assets through that but it’s all shifted to Android the good news I think is a Android and IOs have a lot of overlap in terms of monetization  to to push it further we try to go to singapore and india as a really good proxy for for the us we thought cultural sensitivities in india wouldn’t work because we have me odd h ored clients  but they work it works well and it’s a tenth of the cost domestically  if you move from IOs to Android Android to international you’ll find me really good low cost mechanisms for for doing a b testing got it and what about reactivation campaigns in my experience I’ve never seen  huge percentage of mobile ad budget  kind of dedicated to reactivation campaigns but in this new world of idfa deprecation on IOs  how our reactivation campaigns impacted and what’s your outlook with respect to budget.

You can’t do it. It’s dead. You’ll have again you have 48 hours to determine the viability of a cons er and then you can create me model that says these people are  them but their device ids are gone you can broad spray to the universe to your point it’s kind of  eating your young right you bring back in the same people it didn’t value it didn’t enhance your look-alike audiences or your targeting in any way  people didn’t do it very often I mean they did as much as they could which was maybe 10 of a portfolio yeah but sustained in the future not on IOs what do you guys see happening to the retargeting focused  businesses companies remerge well al there’s a lot of hiking points right these guys are going to pivot and be smart and figure out what to do next I certainly wouldn’t wouldn’t count anyone out here’s the good news because I think the point earlier was a strong one we’re all on the boat together yeah  there’s no advantage and disadvantage this is a fundamental shift that hasn’t happened since mobile app advertising came out there’s going to be significant winners and losers and it’s going to happen in the next 12 months and  to the extent that people can invest now to  put together these pieces that we’re talking about it will provide for a long-term sustained advantage it’s clear privacy isn’t going to go away it’s unclear how google will react my per nal opinion is they won’t do any of this unless forced to by government why would they right they’re the world’s best at delivering per nalized ads why would they d b down their their advertising marketplace it wouldn’t wouldn’t make sense unless countries force them  with with no alternative to do I think IOs is in a really unique place how all of these guys react there’s a really  interesting post today on a mobile dev memo  about what could happen and and the fact that facebook’s I would believe natural reaction would be build up the walled garden higher they’ve got more first party data than anyone and then as you ultimately go to the metaverse and or html5 games right didn’t the whole thing begin with cityville back in the day as an html5 app  why not just bypass the app store and talk in total build up the wall garden around your user base and skip it  I that’s the way people are going to react because they’re going to have to unless the  article that came out yesterday about the senate  initiating a bill to side load games which hey let’s go that would be awe me  because then I think it it shakes up the entire industry in a meaningful way yeah one I heard a really interesting take recently I I was interviewing a a thief khan who used to be a very senior per n at  facebook ads and  he had a very different take as far as I I keep thinking kind of through the lens of  how will  facebook lives off ads how is it going to reshape its business model but  from meone that had been on the inside there his take was that he he thinks that this is just a catalyst for facebook to de-prioritize the ads part of its business there’s  many other large-scale products that they’ve been  brewing there over the last decade or  and he strongly believes this will be a catalyst for just them basically diversifying their revenue streams and just saying rather than trying to fight and and shoehorn in a new working  model for ad revenue just to make it a less crucial part.

Right okay and maybe one last follow-up question as far as this specific topic is concerned but what do you guys think about the cpg style attribution analysis or things  mixed media modeling  how much are you guys looking at that how important does that type of approach become in this new world?

It’s foundational it sucks it’s less efficient and it’s what we have  I think it’s  if it worked we wouldn’t have invented all
this other stuff right you would have left it there  but if if constrained if nothing material changes and this is what you have then you’re gonna have to  package people up in ways that are clustered them a nielsen cluster and it won’t target the individual and it will be substantially less efficient and that’s the future  it will work on  avod it’ll work on ip-enabled tv  for the sustained long-term viability and expansion of the industry in terms of performance advertising it’s it’s a very likely path to go forward okay yeah it’s a I it’s weird I I have a different back  I went to film school I don’t have  a I took  literally one the bare minim math classes are  one math class in a trailer  during s mer s mer break one year I don’t have the stats background but I’ve built a lot of models in mobile but I guess I’ve I I’ve approached it always more from first principles of just  what’s the right way to model for this kind of product  I think that a lot of the stuff. 

We’ve ended up at has overlap  we I talked a lot about our approach to modeling earlier  we’ll rehash that but has a lot of overlap with media mix  approaches  but definitely I think that’s crucial of just  yeah how are you gonna measure the contributions of of each individual channel when you can when you can’t  measure them directly yeah well I think that part of that’s addressed by IOs 15 right I think that’ll be that’ll be very interesting but it doesn’t get you to people can you elaborate on that brain  what  because I just read it last night in terms of what it’s doing but my limited  please don’t harsh on me if I get this wrong understanding of this is that  self-attributing networks  facebook and google have always had an advantage in terms of how they treat traffic  because there’s no arbiter of the real attribution cross network they were able to claim view through conversions that may have been higher than click conversions  in IOs 14 my 15 my understanding is that they will be offering cross network attribution as a as a company  it will normalize the way in which traffic looks across all of the platforms  the networks themselves it doesn’t come anywhere close to addressing what happens at the individual level but at least it’ll put the networks on parity  instead of saying x network is doing better than y you’ll be able to actually see from IOs from apple’s view is that is that true  that’s a if if accurate and if that’s coming out which I understand to be in september would be a meaningful step forward  just not at the deterministic level. 

We will see it one month but  regarding the media mix modeling  right now I’m more using  incrementality tests  to kind of  get me level of me sense  what’s working what which channel is contributing to what is al  a pretty good way how to go okay I mean  to be fair I think we’ve all been wonderfully lazy  with the last whatever is the last click let’s

take it in in the new we’ve had many clients that do paid search and facebook right facebook is often treated  top of
funnel aided  aided branding when they turn off facebook page search efficiency tanks  there’s a logical value in running display and how that impacts other channels of course we’re all going to be forced to figure out the components and lay out an attribution model that makes sense  to the to the point earlier it’s hard to do we’re the pl bers and  we gotta we gotta fix the pl bing all right and this is the beauty part of the industry there you go yeah maybe shifting now to  I know you guys are marketers but when you think about product and from the marketing lens what can product do to help improve performance of games do you guys have any thoughts on that yes that’s my earlier point onboarding flows early monetization signals a deep understanding of your per nas  who are you building for what do they read watch and listen to how are they motivated in the external world which will allow you to cluster creative and interests that kind of core marketing dna will be very important moving forward right and just for our audience Brian when when you say  the rea n why the future initial onboarding experience would be  important is because we have that initial whatever 24 hour windows that we can if we see a monetization event then we can actually get that attributed. 

That’s why you’re saying that’s important but an example would be you put a campaign against people that  to kill each other a
campaign that people  to relax a campaign of people who  to complete levels the calls to action the look and feel those things are all dialed in accordingly those campaigns would naturally accelerate or decelerate and the algorithms will be able to cluster behavior of people who click through that style based on more macro components yeah think about what I read watch listen to the websites I log into all of that first party data that resides at the networks can get clustered into a stack that stack is sustainable which is my definition of how cpg advertising kind of evolves in this right  it’s a little bit more complicated than when I when I initially suggested it any other ideas guys yeah I was going to say I’ve been having I mean we always work pretty closely with the product teams but these last couple of months we’ve been having more meetings with the product teams than than ever and part of this is  to try to try to get on the same page and and  one we have me extreme cases  one of our clients  they have a app that’s in the home renovation space and they monetize off of basically  selling leads for refi loans and  it’s a it’s talk about a needle in a haystack  that  in that case it’s very few users but those users representing  massive amounts of money because it’s percentage of home value  it’s  working with them to back out  okay cool we realize maybe this is one out of x
amount of users it’s a very high n ber but how do we get a day one high vol e data signal that correlates to that because we still got to make decisions off of this that’s it’s kind of a silver lining because I think it’s making marketing teams and product teams starting to work together a little more closely if the org’s not too dysfunctional  but definitely  this is time to make friends with your product team as a marketer and  work on that strategy together identify the events map out the user funnel together figure out if there’s things you can change to to  track mething earlier on that’s indicative of that eventual late funnel event.

In the past it was optional to be very good friends with product now it’s a must definitely and to get the sense of  the day zero to day one whatever you can get and as many events and data you can get that’s that’s definitely helpful for the campaigns and the whole UA front all right  guys I know we’re running out of time but maybe as the last topic maybe quickly we can go over who are likely winners and losers from all this fallout and I know one of the companies that we mentioned was unity  unity seems  they’ve been a big winner although if probabilistic is killed off we’ll see but I don’t know any other thoughts in terms of winners and losers. I mean the obvious is is google right but I’m not really breaking any breaking any news there.

Yeah what do you guys think it’s definitely it’s it’s an hour to shine for all of the small ad networks that were  having trouble getting their emails answered  a year ago

That’s true yeah I mean ironically I think for people who have large first-party data reservoirs right the walls they can build around them yeah large companies are the obvious winners the obvious losers are small to medi -sized businesses let’s be clear that means people spending less than a million a month are going to be negatively impacted those that rely on advertising as their primary mechanism for monetization are going to be negatively impacted right those who don’t have the cash flow to  to roll out an 18-month LTV window and sustain it that’s on the app developer side but I think the very large companies  which is why you see  so much  activity going on are sustainably  advantaged in this change right maybe al other winners would be games that have  mid-core games with broad monetization because as you’re less able to target whales those games will see margin flow to them.

It’s more powerful go ahead right. Owners all warrant to your point about retargeting companies they’ll probably have to shift to mething mething else  anyway let me just throw out a random one netflix I don’t know what netflix games is I have no idea but they have an awful lot of data on an awful lot of people and an ability to  know what you I I think it will be a compelling offering as it comes out yeah well netflix is more  a data company rather than the entertainment company nowadays.. 


Subscribe to Brutally Honest Newsletter


Next