Note: We have a growing community of mobile marketers!

 The Mobile Growth Lab Slack: A community that was a part of our workshop series – The Mobile Growth Lab, is now open to the general public. Join over 200 mobile marketers to discuss challenges and share your expertise. More details are available here: https://mobileuseracquisitionshow.com/slack/

If you’re ready to join the growing community, fill this form: https://forms.gle/cRCYM4gT1tdXgg6u5

PANELISTS Hichem Belouizdad | Carolin Rohte | Lisa Kennelly | Thomas Petit | Shamanth Rao

ABOUT ROCKETSHIP HQ: Website | LinkedIn  | Twitter | YouTube

KEY HIGHLIGHTS

⭐️ Challenges for subscription apps in 2022

🌨 Aggressive paywalls are not the primary way going forward.

🍉 How the advertising channel mix is changing in 2022

🫑 How schemas for subscription apps differ

🧇 Dissecting post backs in iOS 15 and why they are not useful.

🥜 The effectiveness of probabilistic models

🥤 Understanding multiple measurement paradigms

🧉 Incrementality based measurements for subscription apps

🥎 Exploring web flows for subscription apps

🥊 The importance of first party data

🎨 App payment policy changes by Apple

KEY QUOTES

Challenging the pre-ATT playbook

 Thomas: Until ATT showed up there was a playbook that I would say 80 or 90% were following, which is acquisition mainly driven by paid ad monetization – very aggressive, onboarding paywalls. You use the app, there’s a paywall jumping at you every three seconds, which is basically you pay or you stop using. 

I think now with this remix in marketing, there are a lot of people who are starting to question if this is actually the only way? There were a bunch of apps who weren’t doing that, but I think they were the minority. And now they’re looking like people we should be inspired from and should be thinking about.

Channel diversity to get users

Lisa: We still spend a lot on paid, we’re still looking at the channels that have always worked. But we’re also looking at, who’s upstream from us in the customer journey for our specific customer? And how can we get into that conversation? So that’s definitely something we’ve been doing more of.

Alternate measurement models that are being evaluated

Hichem: I can see that marketers are also exploring the top down approach to measurement or evaluating the media mix. And they’re using modelling and incrementality as an alternative to decide how to choose the right budget with the channels.

Schemas that have been effective

Carolin: We have no trials yet, we are currently using only the revenue schema so no conversion value or engagement. We’re not measuring occurrence, meaning we use the six bits solely for revenue. Reasons for that are the 64 values we have are enough for us to map our revenue range plus we don’t have too many events in the app we are measuring yet. So the revenue schema for us works perfectly because we measure the exact value of the user and give a signal to the platform, like how valuable the acquired user is for us. 

How probablistic matching compares with SKAN attribution

Hichem: Probabilistic matching avoids most of the drawbacks that come with SKAN attribution. That’s why a lot of advertisers like this way of working because you get the postbacks in real time. There is no privacy censorship threshold, you can get multiple post backs per user, and each post back has a rich data set. So it is just like the days of IDFA attribution. I believe it allows for more refined optimization. So in theory it should be driving better performance.

How a marketing mix model can help with measurement disruption

Carolin: We are currently working on a marketing mix modeling approach, to estimate the true value, for example, of Facebook, on our business metrics. So the end or the result of this will be aggregated data on an ad level. The outcome will have an impact on channel investment, on overall results. And this will provide us with a macro overview. So no campaign or ad level, that’s not where this is going. But this modeling will hopefully help us to at least, solve the measurement issue of the post ATT world.

What media mix modeling can’t be used for

Thomas: Eventually it did bring a really interesting insight about how those channels have a much bigger impact than we believe. And that was typically the case for higher funnel channels, including YouTube, Tiktok, Pinterest, where impressions do not necessarily convert into install immediately, but they participate to eventually downstream impact. And we did shift a bit more budget on these channels, because it was more than what the last click was saying. So typically, you would assign less to search. But it’s tricky. And it really works at channel level, you can’t make conclusions with that like, I’ve decreased that campaign by 20%, was it incremental or not? Like the media mix model is not gonna answer this, it’s a very strategic high level budget assignment between channels. It’s not tactical execution. 

Thomas’ WIP: A model to match between SKAN and make an analysis towards revenue

Thomas: We’re trying to build a similar matching between the SKAN number we receive and make a regression analysis towards revenue. Like when Facebook is telling me I’m paying $30 for a SKAN trial, can I make the equivalent, that means $60 in revenue, or whatever it is. And there are a number of challenges doing that, it’s really hard to build one because the SKAN value coming from one network and another network, they’re not the same. Like I cannot say okay, a SKAN trial is worth $30. Because then a SKAN trial on this network was 20. On that network it was 45. And two weeks later, the value has changed. So I’m in the middle of building this.

Lisa and her team’s experiments with web-based flows

Lisa: I would say the biggest mistake we made was spending a lot of time building out a web version of the product, which I don’t really think the rest of our marketing organization was sort of set up to support or had the resources to support. So, we see this opportunity for a lot of SEO and a lot of organic growth. And obviously, it’s cheaper on the web, we see this great opportunity, like how do we actually do that, as a company that’s been historically an app company and doesn’t have a lot of web experience? 

And then also, even if we get all that traffic, how do we convert it? And what do we convert it to? Because for a long time, it was like, oh, let’s convert them to just web. It’s cheaper. We’ll save money from the app on Google tax. And now we’re like, let’s get them all to the app. We don’t care. Forget everything on the web. So it’s been quite a journey.

Two popular web-based journeys

Thomas: I see apps both succeeding and failing with two completely different approaches, one being a very direct journey where people would be redirected to a very simple landing page, where the objective is to go to the App Store as soon as possible. And the other takes a much longer route where maybe you don’t have the full product on the web. But you do start onboarding users on the web.

What Yazio learnt about web-based flows through rigorous testing

Carolin: We tested everything, different landing pages, different funnels, funnel length, user flow from creative to the funnel. And our core learnings are that long funnels work better than short ones. Direct funnels work better than going through a landing page, even for those channels that you might think demand more context like Outbrain, native advertising. Our major challenge is obviously that we asked people to purchase a product that they haven’t even downloaded yet. So right now, we are adding a free trial to the web, which we haven’t done until now.

How first party data usage is gaining importance now

Hichem: It said that 85% of US marketers and 75% of Western Europe marketers are saying that increasing their use of first party data is a high priority, but I think it becomes a big priority for everyone. First, because it helps you to comply with global data protection laws. I think it’s a major thing now. It stays in the end of those two connected and gives more control and transparency over what happens with that data. It’s easier to obtain consent for first party data because customers know what you will do with this data.

What we lose out on when trying to maximise first day trial

Thomas: You do see that there’s not a unique correlation of how many people subscribe on day one and how many subs I have at the end of the day. There were channels that were bringing us a lot of what I call the late converters, which very often are people who retain more, invite more their friends, have access to higher value and have higher payment retention and users retention. And today, we’re completely blind to that. So it’s kind of I’m just assuming that those networks are bringing the same after the first day. And I know for sure it’s not right, which is a little bit of an issue. But we have to live with that.

FULL TRANSCRIPT BELOW

Shamanth Rao  

Hi, everyone. Welcome to our webinar on Winning Strategies for Subscription Apps. A huge thanks to our partners at Liftoff and Vungle who are co-hosting this along with us at Rocketship HQ. I’m excited to introduce our panelists.  Carolin Rohte heads performance marketing at Yazio, a leading nutrition app. She began with Yazio as an intern and built the performance marketing team from scratch. 

Our second panelist is Hichem Belouizdad who is a Senior Account Executive CSE at Liftoff and Vungle. He is responsible for driving new business across central, southern and eastern Europe. He has over 10 years of experience in the digital advertising space. 

Next panelist is Lisa Kennelly. Lisa is a marketing and communications professional who is the CMO at Fishbrain, which is the world’s most popular fishing app that caters to more than 14 billion users worldwide who enjoy fishing as a sport. 

Finally, Thomas Petit who is an independent mobile growth consultant working primarily on non-gaming B2C apps. He’s an external consultant for large apps, and is a collaborator for that many app agencies and an advisor to many early stage startups. Very excited to have you guys. This is an all star panel and I’m very thrilled to have you guys today. 

I’m Shamanth, founder and CEO of Rocketship HQ, host of the Mobile User Acquisition Show, check out the podcast mobileuseracquisitionshow.com. 

So let’s jump into today’s webinar. Now that the dust has more than settled on ATT, it’s been a while it’s 2022, what do you see as some of the bigger challenges that are unique to subscription apps In 2022? Carolin, would love for you to kick us off.

Carolin Rohte  

I think there are lots of challenges. Maybe subscription tiredness that might come up or might even become stronger, because everything now can be a subscription in an app. So our job is to make users’ life as easy as possible. For example, expanding payment methods, building a pricing structure to fit customer’s needs, introducing lifetime offers all that stuff, just to make the user as happy as possible. And then I think looking at the broader picture, because missing data accuracy is a huge thing. So for example, looking at blended ROAS, incorporating organic traffic into strategy that’s also nothing really new, but more important than ever, from my perspective. And then the extensive use obviously of first party data because we are not really used to it and we have it now, from Google and Facebook, those would be my three top challenges for 2022.

Shamanth Rao  

Yeah, and we’ll certainly dive into all of those as we go along. But Thomas, would love to get your perspective on what you see as the bigger challenges.

Thomas Petit  

One challenge I see is,

until ATT showed up there was a playbook that I would say 80 or 90% were following, which is acquisition mainly driven by paid ad monetization – very aggressive, onboarding paywalls. You use the app, there’s a paywall jumping at you every three seconds, which is basically you pay or you stop using. 

I think now with this remix in marketing, there are a lot of people who are starting to question if this is actually the only way? There were a bunch of apps who weren’t doing that, but I think they were the minority. And now they’re looking like people we should be inspired from and should be thinking about.

Apps like AllTrails or Strava, apps like Duolingo, built a brand earlier and didn’t fit this other model. I think it’s not about copying Duolingo or Strava. But more like rethinking, there might be another way to rebalance the model; and freemium means freemium, it doesn’t mean you put a paywall to everybody. Can we make value from free users, from virality, and others? But at the price of maybe a little bit less monetization and a little bit less hardcore paid acquisition. A lot of apps, I believe, are thinking about that but need to find balance in the spectrum of, let’s say, monetization aggression. And I think that’s a very interesting strategy challenge for the long term that ATT has pushed us to rethink a little bit. And basically, what I’m saying is that the old playbook is being put in question this year.

Shamanth Rao  

So you can’t be as aggressive at monetization as you used to be?

Thomas Petit  

You can, it’s just one way of doing business.

Shamanth Rao  

There are other paths that have opened up, and we’ll certainly talk about a lot of those other paths. I’m curious, how have you guys seen advertising channel mix change since ATT? How do you see this changing? Lisa, would love to hear from you.

Lisa Kennelly  

So as a result of last year, we realized we had to make some changes, and commit some resources to that. I’ve been doing a lot of work personally within my team on looking at partnerships as a potential distribution source for us. So of course,

we still spend a lot on paid, we’re still looking at the channels that have always worked. But we’re also looking at, who’s upstream from us in the customer journey for our specific customer? And how can we get into that conversation? So that’s definitely something we’ve been doing more of. 

I would also say, there’s more of an emphasis on content, and for us also more of an emphasis on offline marketing than we’ve ever done. Like really a willingness to print up a bunch of flyers with QR codes and putting those in places where people are going to be because what’s interesting, is that there’s a lot more understanding from other stakeholders in the company, and even investors or board members that they really see this too. So maybe there’s more willingness to do that than before, where it was – go give Facebook money, and they spit users out to you.

Shamanth Rao  

Yeah. And you and I talked about this, if you have a QR code, it’s actually more trackable than Facebook is, in some ways, right? Because you’re able to track who’s installing using the QR code, which is an interesting side effect of what you just described.

Lisa Kennelly  

Exactly. We’ll see what the adoption is. How many people actually use a QR code to download an app? I mean, that’s to be seen.

Shamanth Rao  

Yeah. Hichem, you certainly see a number of apps across the subscription space? I’m curious what you are seeing in terms of the channel mix for advertisers?

Hichem Belouizdad  

Yeah, I think that advertisers are trying to figure it out. How they will complete the data, how to optimize to get the complete ROAS or CPA value. And in order to get around these challenges with SKAN, marketers are doing a few different things. For example, shifting the budget to Android to do their A/B tests, because the data is more reliable, at least for now. So spend on Android has increased and the pricing as well. Another option that I’ve seen is to run campaigns on the web. To try and direct users to a landing page to download the mobile app, which gives you website tracking and data. 

I think the influencer campaign as well is getting more and more used than before because you can measure ROI in a more accurate way. And lastly,

I can see that marketers are also exploring the top down approach to measurement or evaluating the media mix. And they’re using modelling and incrementality as an alternative to decide how to choose the right budget with the channels.

Shamanth Rao  

Yeah, so what I’m hearing you guys say is, there’s more exploration of channels. There’s a lot more offline, a lot more on the influencers, web-based flows. And Carolin, you’re nodding so I’m curious if you have anything to add there.

Carolin Rohte  

We’re doing lots of partnership and influencers. We tried TV and out of home for the first time this January, which was of course, something that came out of the whole ATT disruption, so to say. Beyond that, we try to think beyond the duopoly of Facebook and Google adding new channels like programmatic trying new campaign types, Spark ads on Tiktok, adding a web funnel. 

Shamanth Rao  

Sure. So there’s a lot more diversity in terms of distribution channels that you guys see.  I’m also curious, obviously, SKAN still remains a key part of measurement, it’s not the only game in town. I’m also curious, is there anything interesting in terms of the schema that you’re finding has been effective with subscription apps? And because the obvious schema would be to have trials. But is there anything else that you’re seeing that’s interesting? Carolin I know you have some interesting perspectives.

Carolin Rohte  

We have no trials yet, we are currently using only the revenue schema so no conversion value or engagement. We’re not measuring occurrence, meaning we use the six bits solely for revenue. Reasons for that are the 64 values we have are enough for us to map our revenue range plus we don’t have too many events in the app we are measuring yet. So the revenue schema for us works perfectly because we measure the exact value of the user and give a signal to the platform, like how valuable the acquired user is for us.

Shamanth Rao  

That’s definitely interesting, because it’s not a common schema, the vast majority of subscription apps have a trial. That seems to see a very different perspective.

And staying on the theme of SKAN. Obviously, SKAN has changed and evolved ever since it was introduced. So with iOS 15, one of the things that was in the news, to some extent, was that in iOS 15, SKAN is sending post backs to advertisers. Now, technically, that is true, but how has that impacted day to day operations? 

Thomas Petit  

I was very excited when Apple announced this. I was like, fantastic, we have the possibility to get it ourselves. So networks can’t tamper with it, or try to hide it or aggregate it or whatever. And I thought it would be more comparable between networks. And then I didn’t instruct my team to retrieve them at all, basically, because iOS 15 took forever to get adopted. I don’t know why Apple slowed it down. We’re now at 70%, that’s it, but it took six months to get there. 

When the line crossed, 50 or 60% I said, Okay, let’s look at this post back now. It was not what I expected. There’s a bunch of super interesting information, if you’re running media mix modeling and stuff like that, about redownload, about ‘is this a win or an assist’. So that’s extremely valuable, the parameters that come with it, there are two problems. Not only is it simple, because it’s on iOS 15. It’s not the complete SKAN view, it’s just a partial view. But it’s the fact that you receive campaign ID that you have no idea what they are. So it’s like you’re actually spending me campaign ABC, and then I’ve got campaign 123. And I’m like, which A is 1 and which B is 2. 

When you run just a few campaigns on some networks, maybe it’s easy to infer. I’ve got a big one, and a small one. But then you enter a bunch of networks, and Facebook being the biggest of them will run sub campaigns in the background. And that’s why we’re limited to eight or nine campaigns instead of Apple’s limit of 99. Good luck figuring out which campaign is which campaign. To date, I haven’t found any way to match those IDs with my campaign. At scale, you could argue that there is a way to infer but it’s very tricky. And the result of this now, I’m not using it. And it’s a bit of a shame that I’m not, I do believe that these extra parameters bring us something very interesting to look at. But at the moment for me, it’s not practical. So I’m a bit sad that I’m not using them.

Shamanth Rao  

Exactly what I’d say – it’s practically useless. I have spoken to large advertisers who also just have said it’s just not something they are able to use just now. Hichem, I’m curious to hear from you. You guys see a lot of postbacks and data. I’m curious what you’re seeing at your end.

Hichem Belouizdad  

What we see is that for advertisers that are using MMP probabilistic matching, they should be more or less the same as before. Most of our customers continue to leverage MMP attribution. So at least we try to bring to the market the best possible machine learning model that can pose back and MMP developed inputs back to the right performance, which is something very difficult. But I believe that advertisers can adjust to a new measurement model. I wouldn’t be in the advantageous position compared to competitors. So I think it’s the right time now to work on that even if we had a probabilistic matching accurate as from what I see on my side.

Shamanth Rao  

And just speaking of probabilistic models, obviously, that is not SKAN, that’s a non-SKAN measurement paradigm. How effective are you finding this for advertisers? Again, Hichem I would love to hear from you. Since you see a number of advertisers, how effective do you find this to be? How does that answer change with scale? Is it meaningful for advertisers at a small scale? What scale does it change and become meaningful?

Hichem Belouizdad  

Yes, so

probabilistic matching avoids most of the drawbacks that come with SKAN attribution. That’s why a lot of advertisers like this way of working because you get the postback in real time. There is no privacy censorship threshold, you can get multiple post backs per user, and each post back has a rich data set. So it is just like the days of IDFA attribution. I believe it allows for more refined optimization. So in theory it should be driving better performance.

Shamanth Rao  

Yeah, Thomas, you used to look at a number of apps. Curious if you see an increasing adoption with probabilistic channels lately?

Thomas Petit  

It’s hard to say how the market is moving based on the sample. I only see a few apps, I believe, in gaming, there probably was a bit of a shift of money towards networks that use probabilistic attribution, because the richness of this data is so good. Slight shift, because there’s still a lot of inventory on SANS that is very valuable in the subscription space, for a reason that I’m not exactly sure of. But this network has always represented a fairly small total share. 

If you mix big networks like AppLovin, Ad Colony, ironSource and also all DSPs Liftoff, for some reason and subscription mix, they tend to be less, and I haven’t seen a massive shift of money there yet. But I’ve seen a big appetite “We’re not doing any of this, why” and to try. So I believe we’re not yet at this point where the shift has been really significant. And they also have present challenges and to make profitability happen. So maybe it’s early, maybe we need time, maybe the subscription market is a bit slow, but I’m not seeing massive moves off budget. I’m seeing curiosity and questions and tests and “next quarter, let’s do this”. But so far, in my opinion, it hasn’t materialized in a massive shift,

Shamanth Rao  

Yeah. And I think that’s in line with what I’m saying that subscription apps are certainly more willing to test. Certainly, there are a couple of non subscription consumer facing apps that I know are doing very, very well. But again, each product is different. But certainly I do see a greater appetite to test lately.

Thomas Petit  

I see delivery apps, they use these networks a lot. FinTech apps, they use these networks, a lot of gaming, they use these networks. And then you look at the subscription space with education, health, fitness and productivity. They’re not using them that much. And I say, why is that? I’ve seen a few struggle to get performance. Maybe it wasn’t the network’s fault. Maybe it was this specific app. But I do see in general, there’s a gap between gaming, non subscription apps and subscription apps in this. And to me, too, it’s a question I don’t have the answer to – why is the marketing so different for subscription apps?

Shamanth Rao  

Yeah, yeah. Hichem, I don’t know if you want to add anything, or we can go to the next question.

Hichem Belouizdad  

It’s true that the shift is more for the other vertical. It’s what I’ve seen on my side as well. And it’s more difficult for subscription apps. So yeah, that is something we need to work on.

Shamanth Rao  

Yeah, just to switch gears, the other big challenge lately, which I know all of you touched on, at least a little bit was measurement. So we were recently looking at, there’s SKAN there’s probabilistic, there’s Apple ad services and there’s Firebase. If you’re using Google UAC, Google technically has SKAN. But literally every Google account, we looked at, if you look at the SKAN reported metrics, it’s astronomical.  That’s just four different sources right there. How are you making sense of four different sources telling four different stories?  Carolin, you talked about TV, and OOH, Lisa, you talked about the QR code. Obviously, that’s a completely different measurement paradigm. How are you guys making sense of multiple measurement paradigms? Lisa, I would love to hear from you.

Lisa Kennelly  

It’s a shift in terms of internal culture, and being a subscription app. The company’s been around for 10 years with subscriptions for almost that long. To shift away from the mentality of “let’s look at organic versus paid, and compare that ratio,” we say, forget doing that, guys, we’re not doing that anymore. It’s not really helpful. It’s like, we can’t look at this on a channel by channel basis. So I think that’s something that is an evolution. Because as we look at all these different channels, and increase our sort of marketing mix, we constantly still get “okay, and so if we do this, how many registrations will it get us? How many subscribers will it get us?” And sort of saying we can have a benchmark? 

Sure, we should have some goals for that. But, let’s be realistic about how well we can measure that and are we going to see our customer acquisition costs going down across the board, maybe we can segment more into chunks of types of channels. So we do try to do that, like this chunk, we can measure, and this chunk of these partnerships we can measure together versus looking at things individually.

Shamanth Rao  

Cool. Carolin, you mentioned TV & OOH. How were you thinking about measuring all of this? 

Carolin Rohte  

We honestly still compare sources on an ongoing basis. And there are times, when the data between the partner and the mobile measurement partner matches. MMP is still our go to source for everything that goes beyond channel level. And we have a good knowledge on how much deviation is normal for what channel and that helps us. So we know when we are doing good and when we need to wave a red flag. Because, that’s how we handle that stuff. And also, we change our mind and change to a more blended view. So that helped us here as well, not to be that dependent on that number on the ad level or on the ad set level for one specific campaign or anything.

Shamanth Rao  

Sounds like you guys are looking at this as clusters of channels, types of channels, evaluating them differently. Whereas again, in a deterministic paradigm, all channels could be precisely measured. And I know this came up briefly, understanding incrementality. Have you guys used incrementality based measurement? What methodologies have you guys used? What results are you seeing? Carolin, you talked about some of this.

Carolin Rohte  

That’s one of our tries to make sense out of data. We haven’t used incrementality measurement yet. But

we are currently working on a marketing mix modeling approach, to estimate the true value, for example, of Facebook, on our business metrics. So the end or the result of this will be aggregated data on an ad level. The outcome will have an impact on channel investment, on overall results. And this will provide us with a macro overview. So no campaign or ad level, that’s not where this is going. But this modeling will hopefully help us to at least, solve the measurement issue of the post ATT world.

Shamanth Rao  

Yeah. So you’re saying, you can assign value of A to each channel, including offline ones,

Carolin Rohte  

Yeah. We just mash them all together. And then we use linear regression, and then have the model estimate the value for us.

Shamanth Rao  

How reliable are you finding the estimates lately?

Carolin Rohte  

We’re not there yet. It’s a work in progress. 

Shamanth Rao  

We had our first estimation model this week, we are fine tuning but had very promising results. We used Facebook’s Robyn, which seems easier than we originally anticipated it to be. Thomas, I’m curious what you’re noticing in this space.

Thomas Petit  

I’ve seen three different approaches to that. And I have to say they’re all valuable, but tricky to integrate and also in their implementation, they’re a bit difficult. The one that is simple, I work with a partner called Incrmntl. They’re basically an out-of-the-box incrementality analysis. I believe it’s a really good complement to what we get from SKAN and consented cohort and blended metrics. You can’t make it the single source of truth, either. But for me, it’s a really good complement, if you can afford to have one more. So that also adds complexity inside the team. But it does sometimes bring things like “oh, we believe this will happen.” And we will double check “Oh, these models stay the same,” so we would be a little bit more confident. And it’s out of the box. So that’s what I’ve been using.  

The other two were in-house. One is really media mix modeling. And we were doing this before ATT internally. We basically take spend & impressions, and we’re looking at new revenue trends. We try to make assumptions from that. It’s hard to build from scratch, when we did Robyn didn’t exist. So we started from zero. 

And it took us 18 months or something.

Eventually it did bring a really interesting insight about how those channels have a much bigger impact than we believe. And that was typically the case for higher funnel channels, including YouTube, Tiktok, Pinterest, where impressions do not necessarily convert into install immediately, but they participate to eventually downstream impact. And we did shift a bit more budget on these channels, because it was more than what the last click was saying. So typically, you would assign less to search. But it’s tricky. And it really works at channel level, you can’t make conclusions with that like, I’ve decreased that campaign by 20%, was it incremental or not? Like the media mix model is not gonna answer this, it’s a very strategic high level budget assignment between channels. It’s not tactical execution. 

The last one we’re looking at right now, it’s not finished and the more the projects advance, the less I’m confident it’s going to be solid.

We’re trying to build a similar matching between the SKAN number we receive and make a regression analysis towards revenue. Like when Facebook is telling me I’m paying $30 for a SKAN trial, can I make the equivalent, that means $60 in revenue, or whatever it is. And there are a number of challenges doing that, it’s really hard to build one because the SKAN value coming from one network and another network, they’re not the same. Like I cannot say okay, a SKAN trial is worth $30. Because then a SKAN trial on this network was 20. On that network it was 45. And two weeks later, the value has changed. So I’m in the middle of building this. 

If I have to summarize I say it’s good to have an eye on these models to confirm. I mean, as the level of uncertainty rose because that’s what happened last year, basically, there was uncertainty on profitability and on channel return. It provides one more input about am I actually making the right decisions. But it’s not an input you can rely on. So for me it’s a compliment and only a relatively advanced team can afford to have that on top. If you’re operating at a lower level, don’t get into that rabbit hole yourself.

Shamanth Rao  

I agree. Just in terms of finding any correlation between SKAN and revenue, one of the things we tried and one of the challenges is the privacy thresholds. We obviously didn’t know because it’s not documented. A lot of the calculations just don’t help. So certainly a lot of pitfalls here. Just a follow up. Carolin, would love to hear from you since you talked about incrementality and media mix modeling. How does that translate it to reporting because again, as Thomas mentioned, it’s another layer of complexity. So when you’re looking at this data, is that presented with the UA metrics and reports? How does the reporting happen for media mix based sort of results?

Carolin Rohte  

As I said, we’re not there quite yet. But we are not planning to use that data for daily basis reporting, but more of a higher level to see where to shift our money, maybe on a weekly basis, or maybe even on a monthly basis. It’s not something we are planning to really see as a kind of reporting, but more as an overview on how valuable our different resources and channels are.

Shamanth Rao  

Sure, but I think the best we have right now is to show some of the results, alongside SKAN metrics in the table, this is on channel levels, and I think you guys pointed out, it’s not going to work at the campaign level. At the channel level, you’re like, Facebook, here’s the SKAN update, SKAN stat, here is the coefficient we found from incrementality analysis. Here’s the recommended increase in budget according to MMM, so we chose that in different columns in the same tables. 

Just to switch gears, again, we talked a bit about web flows. What have you tried? What are you seeing working? 

Lisa Kennelly  

We’ve been trying to crack the web for a long time. And we’ve had a lot of I would say, false starts in wrong directions.

I would say the biggest mistake we made was spending a lot of time building out a web version of the product, which I don’t really think the rest of our marketing organization was sort of set up to support or had the resources to support. So, we see this opportunity for a lot of SEO and a lot of organic growth. And obviously, it’s cheaper on the web, we see this great opportunity, like how do we actually do that, as a company that’s been historically an app company and doesn’t have a lot of web experience? 

And then also, even if we get all that traffic, how do we convert it? And what do we convert it to? Because for a long time, it was like, oh, let’s convert them to just web. It’s cheaper. We’ll save money from the app on Google tax. And now we’re like, let’s get them all to the app. We don’t care. Forget everything on the web. So it’s been quite a journey. 

Shamanth Rao  

Thomas, what have you seen?

Thomas Petit  

I started this journey in 2017 And I’m still failing at it regularly. A couple of things I learned along the way. One is that people do it for mostly the wrong reason. So when it started it was because “we don’t want to pay the 30% anymore.” That’s totally your wrong reason. And you’ll realize your conversion is more than 30% less of the web’s, you’re losing money. It’s the wrong motivation. But eventually, I keep insisting because one, for me, a big reason to get there is not to circumvent measurement. Okay, I don’t have a scale network, I’m going to go there, because I’m going to be able to track which to some extent is true. But mostly for channel diversification, I want to have more weaponry, they’re not necessarily the same users you’re going to reach. 

Remember, when there was the LAT user, which is now coming to Android, those will not see ads on Facebook and Google because of attribution. So that’s not the case anymore. But I believe there is an audience expansion on participating. Let’s say you only have Facebook app ads and Facebook web ads. For Google, that’s especially true in the sense of UAC can be super efficient for some, but you don’t have too much control about and you don’t have the insight about keywords specific channels, sometimes for like, performance, but also branding reason we want to advertise hardcore on those channels, specifically, as a kind of a paid media play and not your performance. That’s not something you can do on UAC. So having a web journey, in this sense, really enables you to unlock channels that you don’t operate the same way, or maybe don’t even exist, like Pinterest, that shut down their app installed product. 

But anyway, like there’s just a web product and for some specific apps Pinterest is a really interesting valuable channel in terms of the audience you can reach. So for me, it’s really channel diversification and audience expansion that’s the real motivator, but then you get to the execution part and it’s not a walk in the park.

I see apps both succeeding and failing with two completely different approaches, one being a very direct journey where people would be redirected to a very simple landing page, where the objective is to go to the App Store as soon as possible. And the other takes a much longer route where maybe you don’t have the full product on the web. But you do start onboarding users on the web. 

I know a bunch of apps doing that. Noom has been doing that forever, they are pioneers in that. But now a lot of people are trying to copy that. I believe the motivation of onboarding people is to put the payment at the end of it. But that’s not the only reason to do it, you don’t have to drop off at the store page. So it’s true that you can start communicating a bit and and to some extent, even if you don’t sign up and make people pay on the web, you can also raise user intent before they get to the App Store. And I believe that’s a really good way of approaching it. It’s going to be more than one page. But I don’t intend to put friction in there, I have to accept the popups and I have to put in my email and put my credit card. You can make a really nice flow and put the user in a situation after five or 10 screens of I really want this product, because I’m really convinced that it delivers. And for me, that’s a real motivator, besides the audience at the channel is really providing a bit of a different experience than being faced with download that are pushed and so on and bring the user to a state that is more likely that it will continue. It’s a lot of trial and error. 

Sometimes people ask me, okay, what are the best practices about those steps, and I’m like, I’m seeing five apps, so completely different from one to another. So the playbook is being written in front of us. It’s hard to share that practice. I personally find it exciting, because then we have liberty to try stuff. But it makes it challenging because it’s a long cycle. Don’t promise your boss that at the end of the quarter, you’re gonna have results, because that’s not likely. You may get significant growth results for next year. But you’re not delivering by the end of quarter for sure. And if you do, I want to work for you.

Shamanth Rao  

That’s kind of like I said, Noom has a huge, extremely long onboarding. And there’s other apps I found recently that have a three screen onboarding. I recommend useronboard.com for folks that are interested and growth.design, for great inspiration. Carolin, you’ve done a lot of testing on the web, I would love to hear your perspective on what you’re seeing and what’s working.

Carolin Rohte  

Mostly it is similar to what Thomas and Lisa said, because we have been experimenting with a web funnel for one and a half years now. And we’re now at a point where we say the performance is coming close to app install campaigns. So we have done lots of testing and lots of work, we shifted right now around 30% of our budget to the web funnel.

We tested everything, different landing pages, different funnels, funnel length, user flow from creative to the funnel. And our core learnings are that long funnels work better than short ones. Direct funnels work better than going through a landing page, even for those channels that you might think demand more context like Outbrain, native advertising. Our major challenge is obviously that we asked people to purchase a product that they haven’t even downloaded yet. So right now, we are adding a free trial to the web, which we haven’t done until now. And we hope that this shifts the success again.

Shamanth Rao  

Yeah, that’s more testing than a lot of folks that I know, so you have results that are somewhat proven. And you know, one phrase that I’ve heard quite a lot post ATT is understanding how to use first party data. Can you speak to, obviously, what that signifies, but also how that’s become more important lately? And Hichem, would love to hear from you what you’re seeing and taking?

Hichem Belouizdad  

Sure, I read this study about marketers.

It said that 85% of US marketers and 75% of Western Europe marketers are saying that increasing their use of first party data is a high priority, but I think it becomes a big priority for everyone. First, because it helps you to comply with global data protection laws. I think it’s a major thing now. It stays in the end of those two connected and gives more control and transparency over what happens with that data. It’s easier to obtain consent for first party data because customers know what you will do with this data.

Is it for retargeting or for personalization? Is it to resell it to third party data as well? It’s more accurate because you obtain it directly from your customer and prospect. And I believe it’s also cheaper than buying data from third parties. So it has different benefits that I think are more and more important for marketers.

Shamanth Rao  

So from what you’re saying it’s much more of a retention play and a reengagement play. Because of the non-consented users, you’re just not going to be able to retarget them in any way. 

We talked quite a bit about onboarding lately. And I’m curious if you’re seeing a lot of early monetization experience, change post ATT, just because of the way SKAN is constructed, or for any other reasons. Curious if you have any perspectives, Lisa would love to hear from you.

Lisa Kennelly  

I mean, I wouldn’t say the monetization necessarily, but like onboarding, for example, one thing we’re doing, is we’re working with like, Okay, can we drive distribution to partners. And so if that’s the channel and the first place that a potential user hears of us is through like an email from a partner, then we try to make our onboarding flow customized to the partner. So you keep seeing their logo all the time, on your activation emails, it’s all brand with that logo and messaging, so it’s more consistent. So it’s sort of like a little more trust built up throughout the process. So that’s something that we’re doing, I wouldn’t say anything different on the monetization side.

Shamanth Rao  

And that’s via a deep link, where a user clicks on an email deep link, and the user sees an onboarding experience that’s customized to the partner. Carolin, I know you talked quite a bit about some of the experiments you guys have run. Curious what you’re seeing just in terms of the early product changes.

Carolin Rohte  

For us, nothing really has changed, pre and post ATT we have had over 85% of the users purchase in the first few minutes after the download. Yeah, nothing that has changed ever since.

Shamanth Rao  

I do think with subscription apps with free trials in particular a lot of the monetization is front loaded in that way. That certainly is pretty much I would say consistent across all subscription apps, which I think is very favorable.

Thomas Petit  

If I may, here, it’s true that I mean, in some ways, subscription apps were gifted and duped by this change. But the fact that a lot of trials happen, the first day was really a gift, in the sense that we didn’t have to heavily change the experience to actually send this SKAD network some signals, and we could focus on other areas. I’m seeing a lot more work there in gaming, where typically the stronger purchases are coming later. And that’s a problem for us and SKAN network. They’re revisiting what the actual user experience on the first day must be for us to exploit? 

In the sense for many subscriptions, we didn’t have to change that much. I do see a few of them that try to change a little bit of the experience, especially when they don’t have free trials, because the amount of churned subscription is lower. So that makes a weaker signal to optimize for and more threat for privacy threshold. So in one case, let’s specifically add the free trial because of acquisition because the signal is going to be higher volume, not necessarily because the trial was bringing higher revenue, but just for acquisition optimization purposes. But generally speaking, I think in the subscription space, we’re in a decent place there. There’s always improvement in the onboarding for sure. But not necessarily only for SKAdnetwork purposes. 

One thing that I’m a little bit sad with this is it’s a bit of a race towards maximizing the first day trial. And in the past when we add the full flow or when you have it from QR code or whatever,

you do see that there’s not a unique correlation of how many people subscribe on day one and how many subs I have at the end of the day. There were channels that were bringing us a lot of what I call the late converters, which very often are people who retain more, invite more their friends, have access to higher value and have higher payment retention and users retention. And today, we’re completely blind to that. So it’s kind of I’m just assuming that those networks are bringing the same after the first day. And I know for sure it’s not right, which is a little bit of an issue. But we have to live with that.

Shamanth Rao  

Another recent development has been around, is app payment policy changes by Apple. So for folks that may not be aware, can you describe some of these changes, and talk about how this is impacted, or will impact your strategies. Lisa would love to hear from you?

Lisa Kennelly  

I think this is something where, personally, I’m a little burned on it. I got too excited too many times in the past. Oh, this is really going to change things like they’re cracking down on Apple and Google, now we’re gonna have to have alternate payment options, it’s gonna be great. And then nothing happens. And maybe it’s a domino effect. And eventually, it’s all going to collapse. But we see the news about  Spotify, when it’s going to have an alternative payment option. And we’re excited, but I’m like, Okay, let me see what this actually means in practice, because just in terms of our day to day operations and our planning, we’re not changing anything in our active strategies right now, to address that. I’m just very cautious on believing anything until I really see it.

Shamanth Rao  

Yeah. So it’s in the news, but not yet in implementation. Curious if any of you guys have any thoughts on it, about anything you’re testing, trying any opportunities?

Thomas Petit  

I think people are fooling themselves with off-app payment, I surfaced a tweet from six months ago that said developers are gonna freak out when they realize that paying 30% is beneficial. And I believe in many cases, it is. There’s a lot of value in owning the transaction. A lot of the times we’re like, our Stripe users have a much longer lifetime value. Yeah, those are the people that were acquired in another place in a different manner, they are just not comparable. And at the end of the day, it’s nice to iterate, and so on but I believe in most cases, people will stick to in-app payments, because the conversion is much higher, and the hassle is less that you don’t have to worry about tax. 

Only very big apps with big IPs of brand will actually leverage it. It’s not a surprise that Spotify is making this deal with Google, there are two reasons behind it. One, Spotify needs to pay 70%, royalties, licensing fees on what they do. So they literally can’t operate at 30%. I know because I was in a similar business before. And that’s why the platform makes exceptions for this business. Amazon’s got 50% on the App Store, Spotify has a deal with Google. Readers apps, I believe will end up having different rules to the other ones and that’s okay. They just operate a different kind of business. 

I see a lot of subscription apps say, oh look, Netflix is doing that and Spotify is doing that. You’re not in the same business here. They’re in the business of content, you’re in the business of, let’s call it motivation or productivity. Second, you’re not Netflix or Spotify; nobody knows your brand. Nobody wants to give their credit card to that app that has like 100k customers. If Netflix dropped me a Stripe in the app, I’m going to input my card. 

Shamanth Rao  

Guys, this has been great. And I think it’s a good time to start to wrap up. But as we do that, let’s go around the virtual room for a rapid fire question. If you could pick one piece of advice for subscription app marketers to take away from today, What would it be?

Hichem Belouizdad  

We can see that there’s a lot of changes lately. The reporting goals, creative testing, all of that has changed. So I guess you need to ensure that you and your partner have a strong plan and that you continue testing and staying nimble with all of the seniors so good luck to all marketers with the work you have.

Shamanth Rao  

Thank you for bringing your perspective session, Lisa?

Lisa Kennelly  

I guess my takeaway would be to go back up like 30,000 feet. And if you feel stuck, look at who your customer is and what your customer is. Because it’s so easy to get caught in like, oh, what’s everyone else doing with this app? What are different channel opportunities? And remember that you’re you and your app is unique and your customers unique. And if you think about where my customers are and where am I most likely to reach them? That’s going to be probably the safest way to go forward for both the short term and the long term.

Shamanth Rao  

Yeah, that’s good life advice, too. Carolin?

Carolin Rohte  

Despite the fact that everything has gotten a bit more complicated and sure is sometimes frustrating, especially in the last year, it always works to keep the spirit up to stay motivated to Yeah, to just look out for the next best way. And I’m sure there will be one.

Shamanth Rao  

Absolutely. And Thomas?

Thomas Petit  

I’m aligned with the others. Yeah, there’s more uncertainty. But change is also what we do in the industry. I think it’s the interesting part of the challenge. I guess my takeaway here is to think about what Lisa said, don’t try to copy the best practices of others. There are none. The playbook is being rewritten. Nobody knows, we’re trying to figure it out. So copy others, you will copy the errors, follow the best practices it may not apply to you. 

Don’t assume anything, test it for yourself. Trial and error is part of this business. And this is how we find that we just have to accept that. And we also have to accept that my success is not going to be the one for the next guy. So you have to find your own ways. And probably thinking again, about what your customer wants is probably the best way to achieve that. .

Shamanth Rao  

Wonderful, wonderful, thank you for those words. Thank you so much. Hichem, Lisa, Thomas & Carolin. It’s been a wonderful panel, we do have some time for Q&A as from folks in the audience. So please feel free to type out in the chat. If you have questions. We have a couple of minutes to go. 

What was the best UA channel to use for subscription apps from your experience? Which from Facebook, UAC, or AAC would be the best to scale from? Would you suggest running AdMob Ads as well? Have you seen success in Unity AdMob apps? 

I can think we’ve answered that question, Martin. But tell us if that’s not addressed. I think we talked about it in the “channel section” when we talked about the channels and TL:DR as people are trying a lot more SDK networks, Unity, Liftoff and DSPs, suddenly that there’s much more curiosity and appetite. That was a question from Martin a couple of minutes ago. Looks like we haven’t any other questions. 

Thank you and have a great rest of your evening morning, wherever you are in the world today. Thank you, bye. Thanks a lot. Bye.

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!