#65 Apple rolls out SKAdNetwork 4.0 (with Luisa Del Maschio @ Jellyfish)
This week Dan and Dara are joined by Luisa from Jellyfish to help make sense of the recent SKAdNetwork 4.0 release from Apple. It’s a world apart from the web side of things – something that Dan and Dara find out very quickly! Luisa takes them both through what SKAN is, what new features 4.0 brings and the most important differences from web ad measurement to bear in mind.
See more on mobile ad measurement and SKAN 4.0 from Luisa on LinkedIn.
There are lots of acronyms in this episode, so here the lowdown:
- SKAN or SKAdNetwork = StoreKit Ad Network
- MMP = Mobile Measurement Partner
- SDK = Software Development Kit
- ITP = Intelligent Tracking Prevention
- IDFA = Identifier for Advertisers
- ATT = App Tracking Transparency
- GDN = Google Display Network
- DSP = Demand Side Platform
Here’s some further reading on SKAN 4.0:
- SKAdNetwork 4.0 is out, let’s build your strategy from AppsFlyer
- SKAdNetwork 4.0 has launched: Here’s everything you need to know from Adjust
- Adjust introduces our SKAN 4-ready SDK to the market from Adjust
- SKAdNetwork 4.0 now available from Apple
In other news, happy new year – here’s to a jolly good 2023!
Follow Measurelab on LinkedIn.
Intro music composed by Confidential – check out their lo-fi beats on Spotify.
If you’re like what we’re doing here, please show some support and leave a rating on Apple, Spotify, or wherever really.
[00:00:15] Dara: On today’s episode, which comes out on New Year’s Eve, so Happy New Year’s to all of you in advance. We’re joined by Luisa Del Maschio from Jellyfish who’s talking to us about her area of expertise, which is mobile app measurement, and she enlightens Dan and I on what’s new with SKAdNetwork 4.0.
[00:00:33] Daniel: This is a really great conversation all around SKAdNetwork 4.0 and the previous versions. Lots of acronyms, lots of interesting stuff, something we are definitely going to be listened to a couple of times until we get ahead around this stuff. Hope you enjoy, Happy New Year and see you in 2023.
[00:00:48] Dara: Hello and welcome back to The Measure Pod, a podcast for people in the analytics world to talk about all things analytics related. I’m Dara, I’m CEO at Measurelab.
[00:00:59] Daniel: And I’m Dan, I’m an analytics consultant trainer also at Measurelab.
[00:01:01] Dara: We’re also joined today by a guest who is Luisa Del Maschio, who is a paid search director at Jellyfish. So, Luisa, first and foremost, welcome to The Measure Pod, thanks for agreeing to come on and talk to Dan and myself.
[00:01:15] Luisa: Thanks for having me.
[00:01:17] Dara: So what we always do, rather than me doing a really bad job of introducing you, we turn the tables. So, what we usually ask is because we’ve all had a slightly interesting journey into what we do. If you could just kind of give us a little bit of your background, your story, how you got into digital marketing, and then kind of take us up to what you’re doing today.
[00:01:36] Luisa: Absolutely, so as Dara just said, my name’s Luisa, I’m a paid search director and mobile specialist at Jellyfish. So my key role is I head up the partnership with Google app campaigns. So I head up that whole partnership for the agency and help up skill and create best practices for how we should be rolling out our Google app campaigns. But essentially, I got here through university, I did my placement year at Jellyfish and they offered me a job at the end of it and I’ve been here ever since. So I’ve been here for four years, it’s been an amazing journey and this is where I found my love for app analytics and measurement.
[00:02:09] Dara: Amazing, what a nice success story, and actually quite a simple story. Some of the stories we get are quite long-winded and meandering, but that’s quite a nice, straightforward, happy story.
[00:02:20] Luisa: Yeah, 100%. They took the chance on me and I’m hoping they don’t regret it.
[00:02:25] Dara: I’m sure they won’t. So it’s the mobile specialist, it’s the app campaigns that’s kind of the topic of our conversation today. So do you want to give us a little bit of a high level view maybe of so specifically we were thinking of talking about SKAN 4.0, so maybe you could just give us a, a kind of high level overview of what that is.
[00:02:44] Luisa: Yeah, so it’s been the topic on every mobile app marketer’s list at the moment. But I’ll take it back, I’ll go for a bit of the history actually. So it’s been way over a year since Apple shook up the advertising industry by making advertisers ask permission to use the IDFA (Identifier for Advertisers), and it was always a little bit different in that sense to how Google’s rolled these out, like previously, you had to opt in rather than opt out. So this always surprises people but SKAN, the first version of SKAN was actually born in 2018, that’s when it first was released. It’s not actually a new technology that Apple’s released in recent years, but however, the first version was absolutely useless.
[00:03:21] Luisa: So just to give you a bit of a deep dive as to what SKAN or SKAdNetwork is, it’s a privacy centric API operated by Apple. Essentially, it helps ad networks and advertisers measure their ad activity at an aggregated level without compromising the user id. So before this all came about, you could pretty much track everything a little bit like what web used to be like before cookie consent, GDPR really came into play. It was very much like that and then the shock announcement in June 2020 where they mentioned how ATT (App Tracking Transparency) would be coming into play and how SKAN, well the new versions of SKAN would be coming and that didn’t get launched until September 2020.
[00:04:01] Daniel: And that’s, I suppose, where our knowledge is still hazy, but I suppose up to date is the ATT stuff, right? I mean, we are predominantly, well, I speak for me and Dara at least, our history is very much web-focused and especially from an analytics perspective where as you said, everything was very easy, everything’s trackable, you can deploy code very easily in something like GTM. And attribution is just part of the analytics provider, right? And yeah, so the whole StoreKit Ad Network stuff is, it just feels a bit, it’s so alien to a lot of web specialists and I think what something like Google Analytics has done, GA4 specifically is kind of brought these two worlds together, colliding together at colossal speeds, right? So web people and app people, and all of a sudden they’re using the same tool. So I think it’s been a bit of a shell shock both ways to learn both. I suppose that’s where our confusion around this, especially when Apple announced 4.0 coming out and that’s where we really reached out and you were recommended to talk to us. So first of all, thank you so much for coming on to talk to us app-idiots about this kind of stuff.
[00:04:52] Luisa: Not at all, and I think it’s definitely, and what I like to explain to my team is as soon as they put app in front of anything, they’ll get scared. I’ll be like, can you set up an app campaign? They’re like I don’t know how to do it. I’m like, it’s literally the same, it’s literally the same concept. Obviously there is nuances with everything and with Apple it is a little bit complex and it does take a while to actually sit down and unpick it all. And I’ll go into this a little bit later when we talk about the technicalities about SKAN 4.0, but essentially just keep it simple. So when it obviously got announced in June 2020 that Apple will be deepening its privacy stance, and as you’ve probably seen in advertisement, a lot of privacy is, that’s basically their USP, what Apple are really differentiating themselves on.
[00:05:34] Luisa: So SKAN 2.0 launched in September 2020. It was very problematic, a lot of backlash from media partners, and then they gave almost a bit of a leeway for people to get ready and prepare themselves. So they didn’t actually say it was essential to have ATT within the app until 2021. So that was April, 2021 and everyone was waiting for it to drop basically. So they did an update to SKAN in December, which was SKAN 2.1, and then it became essential that people were to, they wanted to capture the IDFA, they had to implement ATT, and that was pushed in 2021.
[00:06:10] Daniel: So with the ATT stuff, that’s obviously the, I say obviously, that’s the thing that Apple released where they needed to get consent before you get into the IDFA. So basically from a web translation perspective, it’s like accessing third party cookies as a user identifier, right? So to get access to that you needed consent, which isn’t new, it’s just new from an app perspective. So that means basically if you did get consent, you could do the old school version of all this, you could just track it, you could do all that stuff. Do you know like if that was well adopted, if there’s still anyone considering that route of things? Or is it all SKAN now?
[00:06:40] Luisa: So, I mean opt-in rates are getting better and eMobi posted a recent stat that 35% globally have opted in, but there is a massive caveat, it’s not as simple as just publishing ATT within the owned app. If you are running app campaigns via, say for example, Facebook, you need that double opt-in from the publisher and the advertiser in order to obtain that IDFA, so that’s where you run into challenges. Say for example, we wanted to implement the ATT prompt, but the publisher that we’re running on doesn’t want to go down that route like Google did there’ll be no way of obtaining that IDFA, which is why it’s not as simplistic as just putting that opt-in message in. But I think Instagram did it in a really great way, you have the choice, you can have a pre-prompt, which warms the user up to understand what they’re going to do for the data and then you can launch the ATT prompt, which is that apps not to track, app allow messaging that you might have seen within your app. Instagram positioned it perfectly in my eyes of like, if you want a personalised and continue to use this platform for the use of creators, then this is what we’ll need. This is essentially what we’ll need to do, so their opt-in rate was very high.
[00:07:52] Dara: Sorry, can I ask a really, probably a really basic question just to clarify that in my own head. So those two levels of consent that are required, one is the publisher so for example, if it’s on Instagram, there needs to be consent for the Instagram platform as a publisher, but then also for the advertiser, is that right? So as a user you need to be opting in to.
[00:08:13] Luisa: To both.
[00:08:13] Dara: To both. Maybe this really is a basic question, but how would you know? Are they two separate prompts that you’d get as a user?
[00:08:19] Luisa: Yeah, so you’ll get the prompt within the Facebook app, and then you’ll get that prompt within, say for example, the ASOS app.
[00:08:26] Daniel: That’s, that’s for retargeting, right? So if you’re doing direct retargeting campaigns, you need to make sure that they’ve been to the app first, prompted, accepted, go to the publisher, prompted accepted, and then you get the IDFA. So I suppose what then, if you’re doing like a UA, I suppose in the context of that, but that’s the User Acquisition side. If you go in kind of prospecting in our terms, if you’re going prospecting for new customers, how does that work? Are you never going to get the IDFA because in a sense they’ve never been to the advertised app before.
[00:08:50] Luisa: Well this is an interesting one, and one that I don’t think is, it’s not common knowledge or they don’t really talk about it much anymore. But essentially that would be it like, If you’ve never been to that app before and you haven’t given them consent, then essentially yes.
[00:09:05] Daniel: So this just kind of reinforcing the idea that in a sense SKAN is the way to do marketing ad measurement and there is very little bits that you might get outside of that. But the reality is that this is going to be, you know, I suppose the one-stop solution for ad measurement. And the rest you can kind of get it maybe, but you would probably tend to ignore because maybe the percentage of all of the data coming through that method is going to be so minuscule that the validity or the sample size is too small to take anything from it, right?
[00:09:32] Luisa: Correct, so the sample size is very, very small. So previous to the launch of SKAN 4.0, you essentially had a 24-hour time limit and you could only get one postback. And so what Apple were saying is you can keep delaying the timer if your events occur sequentially in order, but if you keep delaying the timer, you won’t get that postback. So say for example, so your normal look back window was say 20 days and you wanted to keep delaying that timer until 20 days to get that postback. Facebook or Google wouldn’t be able to optimise their campaigns accordingly because they can’t, physically can’t, it’s impossible. So what media partners like Google and Facebook were saying, in order for our campaigns to work effectively, you need to pick an event that’s going to basically have an indication of lifetime value in the future.
[00:10:18] Luisa: So we would map out six events essentially of what has the higher propensity to become a user or do a purchase essentially. So you have to fit that within that 24-hour window because in order to get the postback in order for the campaigns to optimise accordingly. So it was quite challenging and essentially the data that we were getting back, that one postback wasn’t enough to optimise.
[00:10:42] Dara: Sorry again, I’m going to be the one asking all the really basic questions, but just to zoom out a bit further again. That postback, is that a collection of information? Because I’m trying to get my head around the effect of this timer and why you would want to extend that timer, push it back as far as you can. Is it because you’re waiting for more to happen so you can send a more complete picture back in the postback? Is that right?
[00:11:01] Luisa: Yes, so essentially, so with this one postback you can either formulate it in, if someone does this, this, and this, then this equates to this at the end goal. So you could continue that timer for however long. But in order to actually get good results and optimise your campaign, it was best practice to either look at a signup, for example, would that be an indicator of long-term success? So that’s what we had to differentiate, and it was a conversion value that you would map out within an MMP (Mobile Measurement Partner) of your choice in order to tell the campaign what you’re prioritising. If you think of e-com clients, they’re revenue based, that just wasn’t feasible anymore. You wouldn’t be able to get the revenue or cohort the data or look at what that would look like over a period of time. You had to look at those predictive and those more smaller, softer metrics to map out what that would look like in the long run.
[00:11:51] Dara: That single postback, so the way it used to work with that single postback, was that the only piece of information or collection of information that you would ever get from SKAN, the earlier versions of SKAN?
[00:12:01] Luisa: Yes, and you necessarily wouldn’t get that conversion value if you didn’t hit that install privacy threshold.
[00:12:06] Dara: Yeah, that’s worth drilling into a little bit. So this is a case of the more instals you have, the more information you get, and sadly, vice versa. So if you don’t get enough, then you don’t hit the privacy threshold and you effectively don’t get very good information. Is that the kind of gist of it?
[00:12:22] Luisa: So let’s take Facebook’s algorithm, for example. So Facebook algorithm, you needed at least 128 instals a day, that’s what we predicted in order to get that conversion value back and on Google Ads and some other campaigns, it was slightly smaller threshold. But for Facebook, because of the mechanics on the backend, you had to hit a higher threshold in order to get that conversion value back. So yes, and this is when it’ll make, when we discuss SKAN 4.0, it become a lot clearer why they brought in certain steps because it was this privacy threshold that you had to hit in order for them to feel secure enough to be able to send back that conversion value without being able to track it back to that user.
[00:13:02] Daniel: I have a question about that postback then and sorry if this is a dumb question, but is that first postback the same or instead of the install postback. So I mean, a lot of the ad campaigns are going to be for app instals, right. And I think obviously installing an app is the kind of biggest kind of success metric for a lot of those ads. So is this postback then beyond that as a kind of an event I suppose that happens within an app of one event of which you’re picking as an indicator and you might be able to pass some additional contextual information back using the bits in the inside of it somehow. If postback one is the install like is that one of the same thing?
[00:13:36] Luisa: Oh no, so what essentially happens, so the censorship of privacy happens when the app is downloaded. So that’s why it’s not to do with the amount of conversions you drive or anything. That censorship of that privacy threshold is derived at the point of the install, so when Apple go to assess whether they should postback that conversion value, they’ll look at that privacy threshold in terms of installations. So you will get the install regardless, so a lot of what was happening was you’re just receiving now because they weren’t hitting that privacy threshold and Apple weren’t transparent about what those thresholds were.
[00:14:09] Daniel: So you’d always get the install right? So regardless of thresholds, you’ll always get an install success metric for your ads if you’re driving instals. But there’s this thing about thresholding once you’ve kind of opened the app and you’ve maybe completed level one of a game, for example, that’s maybe your success metric, but you’d only get to do this once, right? You’d only get one postback and you would say, this has happened, or could you send every level, could you send something that this postback with a different set of values within the bits, how would that work?
[00:14:35] Luisa: You get one postback, so that’s all you would get. So what you could do is say for example, they did level one, level two, and that happened within say 48 hours and you were happy with that timer being delayed 48 hours for you then to get your postback. So what you’ll do is map out in your 63 bits, for example, and do forgive me, I’m not a developer, but you would map out level one, level two, and they have to do it in sequence. So before conversion values always had to go up, it always had to go up. So you always had to be a next level, so level one, level two. And then if that user didn’t take the next step, so you mapped out level three by the time that timer had timed out, you would just get that postback on that level two data.
[00:15:16] Daniel: So it’s like just after a set amount of time, you send a postback and in there you pass the information of in a sense how far they got through your kind of thing.
[00:15:25] Luisa: So it’s all done in like a timeframe, so the timer will start, and if you’re delaying that timer, it will delay another 24 hours, another 24 hours. But if no one takes, say for example, level six was where you wanted to stop the timer, but they didn’t go that far, when the timer has timed out you wouldn’t get that if that user didn’t make it that far.
[00:15:45] Daniel: So this is kind of venturing into the world of the SKAN 4.0 changes, right? This is one of the new things, number of postbacks and stuff. So what’s new? What’s new in SKAN 4.0? What can we do now that we couldn’t do before?
[00:15:55] Luisa: Okay, so the first thing is crowd anonymity. So as we spoke before there was these thresholds, right? So previous versions of SKAN introduced the concept of privacy thresholds, but there wasn’t any numerical value or any strategy behind what these were, we just made assumptions as advertisers. So SKAN 4.0 ventures further adding additional tiers to allow more granularity while still preserving individual’s data. So essentially they’ve mapped out this crowd anonymity into three tiers, well, three groups, shall we say, there’s actually four tiers. You’ve got low, medium, and high and as you venture through the different privacy thresholds, the more data, almost like a dangling a carrot, Apple will give you.
[00:16:39] Luisa: So for example, there’s actually four tiers, so tier zero is when null is returned, you’ll receive no install, no insight post-install. And then tier one you’ll get the coarse grain value and have the source idea, etc. which I’ll go into a little bit in detail in a little bit. So that’s what comes with that, so they’re basically saying if you step up the ladder, you’ll get more data and you’ll start to have a deeper understanding the more you hit those privacy thresholds. And then we have the concept of hierarchical conversion value, so this is the introduction of coarse-grained.
[00:17:11] Luisa: So this is when the previous privacy threshold is not met this returns a low, medium, or high value. So really similar to how they’ve structured the crowd anonymity framework. And then you have another value, which is called the fine-grained. And this is really similar to the 63, well, essentially is the 63 bits that we were talking about earlier. And then a step further would be the multiple conversion values. So as mentioned advertisers they only had a single scan postback and that had to capture enough information for them to be able to predict X, Y, Z. Really similar to the Android privacy sandbox that Google announced that you are going to be allowing three postbacks, Apple have followed suit.
[00:17:50] Luisa: So the timer has been removed, thank goodness. However, your first fine-grained conversion value, so this is the 63 bits that we were talking about, will have a 24-hour time limit and then advertisers can receive three SKAN postbacks for a single install each with a specific conversion window given by Apple. So they will decide on the conversion windows, so the first postback will happen in zero to two days, second between three and seven and then third, between 80 and 30. But, you will be able to lock these windows, so you will, as an advertiser, have control a little bit as to when you will get that conversion back, but it has to be between those designated windows that Apple have supplied.
[00:18:32] Daniel: Sorry, a quick question. I say quick, I have no idea if this is going to be a quick question or not, but so does that mean then, so you’ve got these three windows now, zero to two, three to seven, eight to thirty five for these three postbacks, post-install that is. So does that mean that the user of the app has to be active within that window and have done the thing that you want to measure? Because if that’s the case then what if let’s say the first window is like finish level one. The second window is like, I don’t know, complete the first I don’t know, 10 levels and then the third one is complete the whole game. And what if they completed the whole game on day one? Like what happens then if they’ve done the thing that we are measuring in postback three in that first window, what happens there?
[00:19:07] Luisa: Well this is where you’ll need to measure it out effectively and align your events or what you want to map out in line with what you are seeing in your data. You summed it up perfectly where if that user isn’t active anymore, you won’t get that postback.
[00:19:19] Daniel: And so you just have to be, I don’t know, you have to be tactical about the measurement plan around what is it you’re going to use for postback three, because you can’t use something they could feasibly do within five minutes of installing the app right because that’s a bit of a waste.
[00:19:31] Luisa: Exactly and you wouldn’t do that. So for example, we have clients where signup happens in the first day and I think it’s always really, I think this is a really good segway to say, just because the functionality is there, you don’t have to use it.
[00:19:43] Daniel: Yeah, yeah.
[00:19:44] Luisa: So I think it’s like with MMPs, Google Analytics, just because the functionality is there, you don’t have to use it, we do recommend that you map it out just so you get enough, and it could be as simple as, you don’t need to overcomplicate it. There might be something that happens on day 30 or 35 that triggers other high value users. But one thing, these postbacks, there’s no way to link these postbacks together, it’s all in an aggregated view. So you’ll see postback three, however you’ve mapped it out, would be an aggregated view you’ll just see overall this campaign drove X amount of postback three, X amount of postback two, X amount of postback one, you’ll never be able to see it in a linking format and again, that’s because you can’t link it back to the user.
[00:20:23] Dara: I had a quick question, again I’m going to say quick question. It’ll be quick to ask, whether it’s quick answer or not is maybe different. But on the lock-window that you mentioned. Is that, I’m trying to think how to ask this, can you set it dynamically in any way or is it just you select it, say you set it to three days or whatever, that’s it, it’s three days for every user. You can’t have it like combined with something else to say, but if they do this, then set it to four days rather than three. There’s no dynamic element to it you just set it and that’s it.
[00:20:50] Luisa: And that’s why if you are going to lock your windows, you have to make sure it’s consistent and that is probably the day that you’re going to get the most amount of data. So for e-com or marketplace or potentially like aggregators, you might not be able to do that because there’s no way to predict, you get so many purchases on different days, if it is consistent, it will always be. If you locked your window, that’ll be the same across every user. You can’t dynamically change that lock window, if it’s locked, that’s the same. Well, that’s the information that we’ve got so far. It might all change when you’re into the practicalities, but that’s essentially the information that we have so far that once you’ve locked it, that’s going to be the same. So make sure if you are locking it, it is consistent and that’s you know that day is when you’re going to get the amount of data.
[00:21:33] Dara: And on, on your point about just because the functionality is there doesn’t mean that you have to use it. Just to check I’m understanding that right, do you mean that if you like don’t overcomplicate it. So if you’ve got a straightforward app where this really just maybe one action and then, I don’t know maybe I’m oversimplifying, but you know, if there’s a kind of register action and then there’s a completed a game or purchased something or whatever, then you wouldn’t try and create a very convoluted set of steps. You’d just have potentially, you might not even use all of the postbacks, you might just have one. Is that right?
[00:22:03] Luisa: Well essentially, when I say don’t overcomplicate it as like, we’ve been given all these flexibilities, right? Where you’ve got the coarse-grained, you’ve got the fine-grained, and it’s so easy to get so intricate with all these things. But just keep it simple, like for FinTech, for example, look at registrations, look at deposits, look at when the account was funded, and if it doesn’t make sense for your app, it doesn’t make sense for your app. So just don’t try and use all the functionalities that are available all at once. What people will tend to do, what I think people are going to do is they’ll go oh, we can get all this data, but it doesn’t actually link back to the coarse-grained aspect, we definitely recommend like making sure there’s themes with what you are doing in terms of the coarse-grained and fine-grained. Because nine times out of ten you’re only going to get that fine-grained value when you’ve got a high level of privacy. So most advertisers will probably sit in the medium field so that coarse-grained, you’re more likely to only get your coarse-grained values back. So focus on those and make sure when maybe you do hit that higher privacy, you are thinking the fine-grained links to your coarse-grained. So use your coarse-grained as your overall fundamentals and then weave your fine-grained in. Does that make sense?
[00:23:13] Dara: Yeah, it does. And on this kind of coarse-grained and fine-grained concept, is that new to SKAN 4.0? So previously you either met the privacy threshold or you didn’t, so you either got the data or you didn’t. Is this basically now back to those categories that you mentioned. This is saying you’ll get some data if you don’t hit, maybe another way of putting that is there’s basically different privacy thresholds now is that right? Where if you hit the lowest one, you’ll get the lowest amount of data, the next one up, you get a bit more data and so on and so on.
[00:23:40] Luisa: Essentially, Apple are working with the advertisers and Apple are introducing a new way for us just to get more conversion value at a lower threshold. Whereas before, like you said, it was pretty much if you didn’t reach that threshold, you weren’t going to receive that conversion value. So that was the introduction of coarse-grained in order to allow people that hit that low or the medium threshold that you were able to have an understanding of what that campaign had driven essentially.
[00:24:05] Daniel: I think this is maybe the first time Apple’s ever done something that’s benefited an advertiser, right? So uncharacteristic of them.
[00:24:10] Luisa: Definitely, and they’ve also introduced web to app attribution for SKAN as well. So that was announced as part of the updates. However, it’s only available in Safari.
[00:24:21] Daniel: Of course, of course why not? Because what other things would you use? There’s nothing else out there. I have yet another question, but like, let’s say we’re all on board, we’ve got our grains sorted, we’ve got our conversion points sorted, we’ve locked windows. How do we use SKAN? Like how does an advertiser, let’s say I’m going to, I’ve got an app, I want to run some campaigns, driving instals. How do we then use SKAN to measure? Like how do we start getting some data from it?
[00:24:42] Luisa: Okay, so this is where you’re going to have to rely on your MMP, because what we are advising app marketers to do is look at your data, look at what makes sense and prioritise, and then ultimately your MMP is going to be handling the implementation of that. So a lot of MMPs have conversion schemas to make it simplistic and just allows for that centralised source to all be together. But essentially lean on your MMP, work with your finance team to understand and assign values and what you consider is your core events. Bucket your events, bucket your events into three buckets. Low, medium and high, just like the crowd anonymity framework, and that’s essentially how coarse values should be set up. So low, medium, high, what is your priority conversion? What is your medium? What is your low? Again, in the FinTech context, registration, deposit, funded account above X amount.
[00:25:34] Luisa: So essentially I would say, work with developers and your finance team to understand what’s the main priority and then ultimately your MMP will handle most of the integration and the implementation part.
[00:25:45] Daniel: So the MMP is kind of the heavy lifter there. And MMP, that’s the mobile measurement partner right. And that’s the hard thing to get used to, I suppose, coming from a traditionally a web perspective, because I suppose an argument playing devil’s advocate, because I think I know where this will go, an argument there is to saying, well, we’ve got Google Analytics, that measures our marketing campaigns on web and that does our attribution there and we do our reporting from there. GA4 now works on apps too, so that’s the Firebase evolution, Google Analytics 4. So how comes it doesn’t work the same in the app perspective? Why can’t you just use an analytics tool and why do you need a separate tool? Because I can imagine none of these are really free, why would I then need to pay for a tool to do the thing that on a website Google Analytics already does?
[00:26:23] Luisa: It’s a question we get all the time. So as it stands, MMPs have a key role. I think that stems from a bigger question whether, what is the future of MMPs right? But they have a key role with SKAN, they’ve got the solution, they’ve mapped out the methodology, they provide a cohesive, centralised place for us to understand SKAN at a de-dupe level with opt-in data. Google Analytics doesn’t have SKAN functionality at all. So we wouldn’t be able to understand SKAN data in GA. We can see it in aggregate level, and they’ll be modelling, etc. But we will never get that SKAN data and understand and dissect that SKAN data in Google Analytics.
[00:26:58] Daniel: So do you think Google Analytics ever will or Firebase ever will integrate with SKAN, especially if they’re looking to develop their own similar thing down the line. I mean, do you see that ever happening?
[00:27:09] Luisa: Yeah, I do. And actually, when we look at Google app campaigns, it’s really interesting to how they’re measured. So what people didn’t know is that Google Ads has been doing iOS modelling for a while with Google app campaigns on search because of ITP (Intelligent Tracking Prevention), right? They’ve been doing that for ages, so they essentially rolled out this conversion modelling to GDN (Google Display Network) and YouTube, and essentially what they’re rolling out now is something called on-device measurement, which is exactly like customer match. Using that customer email, hashing it and matching it back to that user to get the data back in. So that’s essentially where they’re heading at the moment in terms of Google Ads and matching back and making sure the data’s accurate. But I do see them venturing into that field, it is just complex and messy right now, which is why they’re even saying rely on your MMPs for implementation of SKAN, but they probably will as they learn more around this minefield, but it’s not their expertise.
[00:28:03] Dara: So would you ever, would there ever be a case where you wouldn’t use an MMP, where you would just directly work with SKAN, or is that not even an option? Can you as an advertiser just work directly with SKAN?
[00:28:15] Luisa: You need an MMP, because Snapchat, TikTok, Facebook, all have different dynamics when it comes to SKAN, but they all look at it in different ways and they all have different privacy thresholds. For example, Facebook, they have a dummy campaign on in the background, which is why their privacy threshold is far higher. They all deal with it in different ways, and a way to see that in a centralised view, it just would become impossible.
[00:28:37] Dara: And just sticking with the subject then of MMPs, if I’m new to this world, which I kind of am, and I’m looking for an MMP, what would I be looking for? How would I compare them to figure out which one’s best for me?
[00:28:47] Luisa: It depends on your goals? I couldn’t really say which one I would recommend because they’re all different and they have different USPs in that sense. But typically at Jellyfish we use AppsFlyer, but Kochava has an amazing viewpoint on how they’ve integrated iOS, how they’ve integrated SKAN, for example. They have a lot of flexibility in what you can do within their schemas. Singular, pretty much spearheaded the whole SKAN conversation. So it all depends on your needs, you sometimes get customers who are not investing on iOS, for example and these are expensive, so yeah, it’s hard to recommend one because they have their different weaknesses. But from AppsFlyer we like the fact with their incrementality testing and how they roll that out, and when we haven’t found that within an MMP before, that’s sometimes why we, we tend to stick with them, but I wouldn’t have a recommendation because Kochava and Singular do a fantastic job in SKAN, and they were the first movers and have become quite the leaders in that space.
[00:29:46] Daniel: So Luisa, I’m normally the one that brings everything back to Google Analytics 4 somehow on this podcast. I’m just going to live up to my title that I’ve been given there. But let’s say, if I’m running app install campaigns, I kind of understand SKAN and the need for SKAN and the need that something like Google Analytics won’t be able to track or any analytics tool won’t be able to track through the install process on either app store, right? That’s just a nightmare, full stop. So let’s say you are not driving app install campaigns and you are doing sort of like, I think they’re just called reinstall campaigns, but in a sense, re-engagement campaigns of users that have the app installed, but you want them to re-engage with the app, would we then need SKAN to come into place for that? Or could we go back to the equivalent of UTM tags and Google Analytics, but for the app world, does Google Analytics do anything in terms of that kind of ad measurement?
[00:30:28] Luisa: No, so re-engagement is a bit of a tricky one. The marketing pools are so small, so what you can do, for example, is if you’ve got opted-in data, you can upload device IDs and target those via iOS. They are coming back and the use of being able to do that is being allowed. But just to talk from a Google standpoint, they stopped their, they literally stopped re-engagement campaigns. The marketing pools are so small it is really tricky these days to retarget on iOS.
[00:30:58] Daniel: Wow, so that is purely going to be things like push notifications then, it’s going to be the only way you’re not going to run any kind of paid ads to promote re-engagement then?
[00:31:05] Luisa: There is some documentation recently on how Google Ads is going to start reintroducing them. But yeah, essentially when this did first happen, re-engagement was pretty much no more because of really low opt-ins. And this is also why you’ve got to do a good job of allowing the user to trust you first. So this comes back to the point of when do you implement your ATT prompt, right? Do you do it to begin with or to get that IDFA quickly or do you do it once you’ve built that trust? So it all comes down to again, being strategic and knowing when, testing when to place those prompts.
[00:31:38] Luisa: But with the release of SKAN 4.0, we are going to be able to bring back the use of creative testing within campaigns across our media stack because they made changes to the campaign ID. It’s now changed to source ID, which allows for four digits to come back, which now means that you can start to test. You can tweak the digits as you please, but there’d be an element that we can bring in creative testing back, which is fantastic.
[00:32:02] Dara: What was the problem before? Why could you not do the creative testing previously?
[00:32:08] Luisa: So basically there’s something called hierarchical source IDs. So it was a two digit campaign ID, which would help you understand which campaign it was coming from. So because there was only two IDs, there wasn’t much room for being creative with these IDs. Does that make sense? I don’t want to confuse anyone, but essentially what Apple have done is added two extra digits to open up more data and richness for testing creatives and formats so you can play around with the digits. Whereas before you couldn’t really, because you needed to use those two digits up to identify where that install essentially was coming from.
[00:32:43] Dara: Yeah, no that makes complete sense. I was going to ask you about that because I just jotted down some of the features of SKAN 4.0 and when I got to that hierarchical source and identifiers, I didn’t really know what it was. So I was going to ask you, so you just answered that.
[00:32:54] Luisa: Oh, amazing. So this essentially falls back into, back to the crowd anonymity discussion. The more you go through low, medium, high, the more you’ll get added, so the more data you’ll get added. So the source IDs, essentially you’ll get more of them as you go through and you hit that higher level privacy threshold. So on the higher level, so say you reached the higher level privacy, you would receive more of those source IDs, so digits two, three or four.
[00:33:19] Daniel: You mentioned already that Google have started to talk about doing something similar to this, and it may have even influenced Apple’s hand in this SKAN 4.0. Where do you think this is going to go? Let’s say we had the same conversation in like a year or two years time. Do you think that both Apple and iOS are going to be exactly the same? Do we see this going to web anytime soon? Like what does that picture look like?
[00:33:39] Luisa: It’s a really interesting one in terms of it going to web. I think the next release will probably be around how do we integrate that on Google Chrome? How do we get the web to app attribution on Chrome? That’s everyone’s preferred browser of choice, but we can only do that web to app attribution on Safari. And for a lot of multi-touch clients who are across web and app, that really does play a huge limitation for them. But in terms of where do I see this going. So you’ve got Android privacy rolling out next year, I believe, and I think Android privacy sandbox is going to be worked way more in the advertiser’s favour with actually having to opt-out rather than opt-in. But we don’t even know, like we’ve been having discussions, will SKAN be a thing in, in two years time. Like, would they completely wipe the slate and think of something completely different? They are just building on and building on what they did launch in 2018, and it might completely change. It’s hard to tell where this is going to go. There is logic in their thinking once you start to unpick it and understand it, it’s just literally, it does make logical sense, but it takes a while. It’s not the easiest of things to get your head around.
[00:34:40] Dara: And it sounds like they have listened. So, you know, the previous versions they were pretty harsh, if you were an advertiser, you would’ve seen them as being pretty harsh and I think you even said they were, it was effectively useless at the beginning or close to it. So it does sound like it’s going back into a more kind of position in the middle maybe.
[00:34:57] Luisa: I think ultimately they had to, because I don’t know if you ever saw when it first launched. So, when they first launched SKAN and they announced that ATT would be coming to effect, they said Apple Search Ads won’t be affected. Apple Search Ads, they’re a platform, they’re not an ad network. So effectively they covered themselves, that caused massive uproar, I don’t know also if you’ve seen the rumours that they want to start producing a DSP (Demand Side Platform), etc. If they want to start playing more in this advertising field, they will have to respect the advertisers that ultimately are going to continue to spend money with them. Because you saw the results as soon as ATT got launched, everyone pulled their spend from iOS, it did go back into Apple Search Ads, that revenue boomed for everyone from Google Ads, Facebook, everyone pulled their money from iOS because they didn’t know what to do with it they didn’t know how to interpret the data.
[00:35:45] Daniel: Facebook lost 10 billion, projected to have lost 10 billion in 2021, which was, you know, from ATT it’s like, and that’s just Facebook. And unsurprisingly, it sounds like Apple’s profits didn’t take a hit, those went up.
[00:35:58] Luisa: No, but they didn’t hit their revenue target for the ads, so I don’t know if you’ve seen the recent announcement that they started to introduce new placements on App Store today. Yeah, they’ve introduced two more placements, so they’re continuing to invest in this advertising world, and I think that’s why they have to be like, okay, we’re coming to play now. We’ll starting to do stuff in your favour and which they had done, but it’s still, it’s still not how we would like, but it’s better than nothing.
[00:36:22] Dara: To flip it round, I know maybe it’s a little unfair to ask you a question around kind of like the kind of hardcore privacy question, but I’m going to ask you anyway just to get your kind of thoughts on it. But is there anything from a user point of view, is there anything that’s being kind of taken away because to get that position back into the middle where it’s, I keep saying fairer, but where it’s more towards the advertiser, more suitable for the advertiser, does that come at a cost to the user? Because I know Apple are saying this is still very privacy centric, but do you think there’s something being taken away from the privacy side to kind of please advertisers or are they able to kind of toe the line there?
[00:36:57] Luisa: That’s a really good question and I think you’ll be able to unpick it more when it starts to roll out, whether there is actually any level. I don’t think they’re taking anything away from the users. I think the way that they’ve laid it out and how you anonymise the data and how we’re using the fine-grained and coarse-grained, I think it’ll be very, very hard to get back to that user. I think it’ll be virtually impossible.
[00:37:17] Daniel: Do you see people doing, sort of moving over to SKAN 4.0 now? Like are you seeing the adoption within the industry? Is it strong? Is it getting there or is it still very early?
[00:37:25] Luisa: I think it’s still very, it’s very early days. The first movers can move now, you’ve got Kochava, AppsFlyer already, Branch already saying what you can and can’t do within the studios now, like conversion studios. But it’s still very early days, nothing happens very quickly. So you still are able to use like SKAN 3.0, but ultimately in order to unlock all the new features you will have to update your Safari, your app, etc. and that user will have to go on iOS 16.
[00:37:49] Daniel: Yeah, I was just thinking that all of the bits that need to be in place to use it, so the user has to update their device to 16 plus, the app has to be updated with the latest SDKs (Software Development Kits) from the MMP, the MMP has to support it and the publisher has to, or sorry, the network has to support SKAN 4.0 as well.
[00:38:05] Luisa: It does take a while and that full adoption won’t happen straight away. And obviously this is a massive announcement and people are going to have to go back to the drawing board essentially of how they start mapping these out and valuing them.
[00:38:16] Daniel: Well it’s the same thing with Google Analytics I think. What they do is they introduce a new version every couple of years just to keep people like us busy. I think the same with SKAN it’s just to keep us working and busy, right?
[00:38:26] Luisa: Literally, but there’s loads of announcements coming. It’s great being in a position where you work with Google and Apple because you see the way that they’re going. But Apple is going to bring out benchmarking in the app analytics because at the moment that app analytics isn’t hugely insightful. So they’re going to be bringing out benchmarking around crash rates, re-downloads, category ranks, etc.
[00:38:47] Dara: I can see why you said at the beginning that you find this a really exciting space to work in. I’m not sure I could keep up, but you seem to really revel in the kind of slightly complicated nature of apps compared to maybe the more straightforward world of web analytics.
[00:39:00] Luisa: Honestly, it’s a minefield but it is exciting, it is exciting times and Apple is coming to play and I think it’s welcoming Apple into the mix. It’s going to change the game and I think, yeah, they’re definitely making a massive standpoint.
[00:39:12] Dara: Well, thank you again for coming on The Measure Pod to enlighten us on this convoluted, convoluted world. I’ve definitely learned quite a lot. Before we let you go, we’ve got two more questions for you. The first one is, what do you like to do when you are not helping people understand mobile app measurement? So what do you do outside of work to wind down?
[00:39:34] Luisa: So I’m actually, I’m a gymnast, so I do gymnastics Mondays, Wednesdays, and Sundays. So I said to myself, this year you’re going to get back into it. So I was a gymnast for like 10 years and then I could try to get back into it and I’ve tried to learn a new niche within it. So that’s what I do in my free time.
[00:39:51] Dara: Wow, that’s really impressive. So were you like a competitive gymnast?
[00:39:55] Luisa: I competed in Team Gym, so a different discipline. But yeah, I competed in Team Gym for a good amount of years.
[00:40:02] Dara: Amazing, brilliant. What a contrast from mobile ads.
[00:40:08] Luisa: I know, I know.
[00:40:10] Dara: Amazing, final question. Where can people find out a bit more about you or get in touch with you if you’re happy for them to do that?
[00:40:17] Luisa: Absolutely, via LinkedIn.
[00:40:18] Daniel: I’ll share a link in the show notes, so there we go, we can point people that way.
[00:40:22] Luisa: Yes, perfect. Yeah, LinkedIn’s the best way that’d be perfect.
[00:40:25] Daniel: Great, thank you so much for coming on. I think I’m going to have to listen back to this one a couple of times and take it all in, but thank you for humouring us and taking us as lowly web people through this crazy world of app.
[00:40:35] Luisa: Oh, you’re so welcome. And nothing is a dumb question in the world of app, honestly, nothing. We are all figuring it out together.
[00:40:42] Dara: I don’t know, I held back some of my really dumb questions.
[00:40:47] Luisa: You shouldn’t have.
Dara: That’s it for this week, to hear more from me and Dan on GA4 and other analytics related topics, all our previous episodes are available in our archive at measurelab.co.uk/podcast. Or you can simply use whatever app you’re using right now to listen to this, to go back and listen to previous episode.
Daniel: And if you want to suggest a topic for something me and Dara should be talking about, or if you want to suggest a guest who we should be talking to, there’s a Google Form in the show notes that you can fill out and leave us a note. Or alternatively, you can just email us at firstname.lastname@example.org to get in touch with us both directly.
Dara: Our theme is from Confidential, you can find a link to their music in the show notes. So on behalf of Dan and I, thanks for listening. See you next time.