In this episode, Guido Crego, VP of Product at Jampp interviewed experts from Singular, Dataseat, and Bamboo to hear their take on the upcoming privacy modifications in iOS 14.
App Marketers Unplugged is an event series dedicated to discussing the latest trends and challenges in our industry. This year, we’re introducing curated video podcasts with a new selection of industry experts.
In this episode, Guido Crego, VP of Product at Jampp interviewed experts from Singular, Dataseat, and Bamboo to hear their take on the upcoming privacy modifications in iOS 14.
Guido: Why do you think that this change in iOS 14 is such a big deal compared to other initiatives regarding user privacy like GDPR and CCPA?
Gadi: The first thing that comes to mind is that GDPR is an opt-out mechanism and advertisers have more tools to customize the consent. In some cases, they can claim legitimate use. Now, Apple's is basically forcing you to be in an opt-out state by default, so you now need to get people to opt in, no matter what. Also, Apple has more tools than the European legislation. They own the platform so they can enforce it much faster and with technology, unlike the courts.
David: The response to GDPR and CCPA has been a lot of people updating their terms & conditions and listing which adtech partners they send data to. That's really been the most noticeable change. IDFA is the consistent identifier that a hundred billion dollar industry operates off, so if I have to draw a comparison to IDFA I'd compare it to ITP, what Apple did with Safari: Intelligent Tracking Prevention 1.0 then 2.0. That was pretty similar.
Daniel: Another thing to consider is that these changes impact billions of people and a lot of market cap, and so any change, whether big or small, feels way more consequential—because it is.
Guido: When the iOS 14 news first came out, the industry kind of panicked because it was a major disruption and we didn't have much time to prepare. Since then, Apple has pushed these changes to early next year. How do you think companies need to leverage this time?
David: We spent a long time studying SKAdNetwork. The biggest limitation from a “running campaigns” perspective, is the integer. We are limited to 100 campaign variables in that postback to Apple, compared to the millions of variables that we have today. So advertisers need to think about the smartest way to use those integers: what are the most valuable parameters that lead to performance that you want to get back? If you study the documentation, the postback that you get from Apple includes the publisher ID, that to us was like “excellent!”
My advice for marketers is to work with their partners to get an understanding of it, and also to identify which of their partners haven't got a clue what they're talking about.
There aren't many people with thought leadership that's driving this forward.
Daniel: What we are advocating for, at least in the interim, is really making sure that our clients have the infrastructure within their products.
Basically making sure that you cohort your users based on IDFV, so that internally you understand what customer behavior looks like. We also try to identify early customer lifecycle activities that correlate with that longer-term value so that we can still use those things to bid on Facebook and Google.
Also, I think we're going to move into this “trust but verify”, where you really need to have good product analytics infrastructure to see what’s happening after SKAdNetwork is not giving you anything.
Gadi: Another very important, yet very limited integer is the conversion value. With SKAdNetwork, we can only describe the value of the customer within a short period of time and with a very small amount of information. We have a number between 0 and 63 that needs to designate the value of the customer. Companies need to figure this out ASAP and the thing is they can't because no one is live with a single SKAdNetwork campaign yet.
The second thing is what's going to be your reporting infrastructure. With SKAdNetwork a lot of the data you're getting is just numbers, so you need to translate: what is conversion value number 17? Maybe it means someone made a purchase and they signed up. What is campaign ID 95? In some networks, it may mean a combination of the creative and the sub-campaign, in Facebook, it may mean something else, so there's a lot of mapping and translation you'll need to do.
My advice is: start thinking about how your UA is going to look like the day after IDFA goes away and what your dashboard is going to look like.
One last thing I want to say is, the folks at Jampp did a good job working with a lot of networks to get ready and you’ve been very proactive. So I really appreciate that.
Guido: Right, and it’s really difficult to test SKAdNetwork today because Apple only sends you postbacks after a certain number of conversions happen, and nobody knows what that threshold is, so you need to run real traffic in a high volume to actually receive something. Considering this, do you see advertisers actually getting ready for this?
Daniel: I think people are looking to Facebook, Google, Singular, Appsflyer, and Adjust to give them guidance and be the leaders through this. Like Gadi said, you can't test anything and there's no data. There's very little information, so the extra time was helpful, but it's still kind of a mess.
David: I agree with Daniel, I also think that there's a part of human nature which gets the most done when you're panicking. So I actually think the early deadline was good because we all scrambled and we already got our SKAdNetwork tracking in place. I do agree people are turning to Google and Facebook, but honestly other than Facebook saying “this is what we're doing for our campaigns”, there isn't much thought leadership or advice coming from them.
Gadi: Some of the bigger organizations already have their data science team coming up with models and experimenting with conversion values. I've seen some of our more sophisticated customers making multiple bets trying to say: “ok, can I predict the lifetime value? What if I can only report after one day or what if I get a few more days of conversion values?” There are slight variations between the really super sophisticated folks starting to research what they need to do, but I would agree most people are looking to be guided in this process because it's difficult.
Daniel: Do you imagine that this change will have a rich-get-richer dynamic, where folks who have the infrastructure, tools, and teams will win in a massive way, and everyone else will fall behind, or will it be democratizing?
Gadi: I guess there are two opposing forces: the bigger customers with the more sophisticated data science teams and analysis capabilities, they will do things that maybe cost more money and give them an advantage; the other opposing force is that Google and Facebook also have a lot of resources and their job is to make it so that everybody can run successful ads on their platforms.
The other debate is: is this really good for Facebook and Google and damaging all the other ad networks? And there's two views on that. Some networks say “this may bring Facebook down to our level because Apple won’t let them use all their spy mechanisms to look into what users do anymore”. The other side of that is that Facebook might still have an advantage if they can pull off something like advanced matching and start using emails. That’s an advantage for Facebook and Google because they do have my email. It's hard to tell if you only benefit the rich people or if it democratizes things...
Daniel: Like you said, it does democratize things with Facebook in the short term, but in the long term I can see Instagram or Facebook shops, where they basically pull even more of the commerce infrastructure of mobile onto their platform. That's bad for all advertisers and ad networks, but very good value for Facebook.
Gadi: Yeah, and also Apple's motivation is privacy, not necessarily justice in the ad ecosystem. Their number one objective, if I read it correctly, is to give consumers control and privacy.
David: I agree that the headline is “Apple: privacy first”, but it brings the question, once you think about the Apple toggle, does this allow Apple personalized advertising? Because that to me is a little bit smelly. Is there an Apple kick-ass advertising product coming down the road that's there to eat everyone's lunch?
Gadi: My take and I may be completely off here, and one day people will see this recording and make fun of me, but I think what you are seeing is strictly a lack of coordination between different teams, so that Apple Search Ads engine is running on old attribution technology that they had on the iAd days and they just didn't adapt it yet to SKAdNetwork. I think that will be solved.
I don't think they're worried about the impact on their ad business as much. The data in Apple Search Ads is solid, they can target anyone they want and they know the downloads because they own the App Store. So it doesn't track post-install, but they have perfect attribution. I don't know if it disrupts their business that much to play by the same rules and enable SKAdNetwork on their products.
Guido: To summarize, I think that we can all agree that this is a big reset. The rich today are rich because they have a lot of behavioral data stored on their user graph and that's going to lose a lot of value. But in any crisis there are opportunities for new business models, so what do you think these changes will open in terms of business models and new opportunities to do things differently?
David: I think anyone focused on driving performance with contextual variables and how to convert that into the SKAdNetwork limitations will survive and thrive.
Daniel: This is an opportunity for marketers to become much more cross-functional than they are today. We've gone through almost a decade of this signal from the big ad networks, DSPs and supply side. We're now moving into this verify era where marketers should really understand how to slice and dice their own data, their own cohorts, their own behavioral analytics and use that information to build better audience segments, value props, messaging, and better creatives. So I’m excited about that change. I think it's going to force marketers to really understand their customers and what is driving their customer behavior in a much more fundamental way.
Gadi: I have to say that as much as there is a lot of sophistication and new vendors coming up promising all sorts of things, simplicity wins and there's a chance that SKAdNetworks get a bit better or maybe people learn how to work with what they have right now without super crazy tech.
There is a lot of room for innovation and trying to have better prediction LTV and cram that into SKAdNetwork. Our job is to find ways to make marketing effective. The geek side of me has got to say that our roadmap has never been more exciting. We're iterating a lot of Machine Learning and AI, doing a lot of cool experiments, and my team is excited about that.
I think we will see the reporting tools and measurement tools just become a lot more sophisticated. We'll see new signals coming in, but the SKAdNetwork will probably have a pretty major part in all this, so I guess the opportunities are opportunities: more innovation, AI, and more sophistication.
Guido: We all agree that SKAdNetwork is here to stay, but it's important to remember that we are actually dealing with the 2.0 version. If we daydreamed a little bit and made a wishlist for the hypothetical SKAdNetwork 3.0, what features or improvements do you think are key?
Daniel: For me, the easiest win would be just more transparency around how all the arbitrary stuff works. I understand that they are randomizing so people can’t fingerprint and reverse engineer, but we do need a little bit more information. So more transparency from them would be the first thing, and I also think way more configuration on attribution windows settings, because we work across so many different types of mobile products and all of them have very different customer journeys and behaviors.
David: I have two asks, but being realistic I don't think we’ll get them. The first one is to have a string instead of an integer. If it was a string, that could be an impression ID. It would allow us to know that this conversion came from that impression, and that allows for optimization. And if I can't have a string, then can we have more than 100, please? That's my first request.
My second is the delay. I think they're going to the end's degree to protect privacy, but, you know, I read this pop-up message saying: “you're being tracked across all these other sites and apps owned by other companies.” 99% of it is just attribution and optimization to make advertising effective, it’s not big brother.
So I would like a string and to remove the delay. It just reduces the effectiveness of advertising which leads to irrelevant ads being served to consumers, and I think that will annoy them more than allowing basic attribution optimization.
Gadi: It’s about finding the balance between privacy and functionality. The first thing I’d suggest is to support more conversion values. However, you can't just have real-time postbacks because then it will be easy to identify and Apple wouldn't approve it. So maybe a compromise is to send the postback once in a couple of days, something that at least enables us to build cohorts. Right now everybody is struggling to build LTV prediction models just because SKAdNetwork is so limited.
The second thing is supporting more granularity with the same token. Why did they determine 100? Is it really the most probable threshold? Could you make it 200 for some advertisers? Could you make it dynamic?
The best thing that happened with SKAdNetwork 2.0 is that they added a version identifier that suggests there will be a 3.0 version. If they're going to iterate, I’m hoping that they will add more granularity, whilst still maintaining privacy.
Lastly, if someone opts-in, let the SKAdNetwork postback contain the user's IDFV so we can deduplicate between SKAdNetwork and not. If conversions came from the SKAdNetwork and that user has opted in, you’d also see an IDFA conversion, and without a consistent identifier between the two, you can’t deduplicate.
Guido: Do you think it's possible that we’ll see any improvements by the time Apple starts enforcing this early next year, or we should wait for iOS 15? or 16…?
Gadi: I think that in the last release they definitely put more effort into SKAdNetwork: they made videos, they opened a support team that accepts your registrations, they’re asking for feedback from developers and even vendors, so it seems like they're serious about it. The time between 1.0 and 2.0 was like two years. Hopefully, the time between 2.0 and 3.0 is going to be shorter.
Guido: So what about Android? Do you think Android will respond in the same way and how and when. What do you think about that?
Daniel: I could see them responding to actual customer behavior, so if there's a huge loss of market share or something like an actual tangible business metric for them, but who knows how reactive they would be.
David: They've also got Google Play Referrer. So you could actually kill Google Advertiser ID (GAID) and you would still be able, you know, Gadi's business would still be able to do deterministic attribution with Google Play Referrer. That's what you were referring to in one of the improvements: for Apple to allow these additional parameters, to track mobile web—that's like Google Play Referrer.
Gadi: Google Play Referrer is a dream. If Apple could do that, that'd be amazing.
David: Yes, if they got rid of GAID, it would still affect retargeting and data platforms, but while the IDFA changes make it impossible to attribute an install granularly, Google Play Referrer does allow that, and that's the first thing. The second thing that makes it less likely is that Apple doesn’t have a huge advertising business, but Google does. Google makes billions of dollars from all these ad exchanges buying their inventory just as DSPs do. If they followed Apple’s lead, it would have an immediate billion dollar effect on their own revenue.
I think Apple’s biggest issue with IDFA is the fact that it wasn’t just used for attribution, it was used under the disguise of suppression lists to be sent to thousands of ad networks.
How do you remove the downside, which is the abuse of deterministic consistent identifiers, whilst keeping the good side? And the good side is relevant ads and good ROAS for advertisers and the buoyant industry that employs us all.
Gadi: There's all sorts of theories that Google will create this monolithic one system that will control everything and I think that's actually going to open them up to even more antitrust and monopoly issues. They are already in very dangerous waters where they control way too many assets. They're making money from ads, not from measurement.
What they are probably doing is waiting to see how well it goes with Apple. Google only did the depraction of third party cookies after years of ITP being out. ITP kept adding more restrictions until the point where Google finally acted, so I’m not sure why they would have to rush to make changes on Android. Android is still a massive platform, so maybe they won’t have any issues.
Do you think there's a chance some more spending will move to Android? What if it becomes easier to acquire customers and grow on Android than it is on iOS? I’m curious about what will happen to our customers’ budgets, like what will happen in a year from now.
Daniel: Or 5 years from now, if every person who has any amount of wealth moves to iOS because they want “privacy, security” on ads like their core foundational future. A lot of things could unfold.
If you have questions, or if you just want to join the discussion, be sure to join the Mobile Attribution Privacy Slack Group. It’s free, it’s open for everybody and it’s a good resource for marketers to learn more about implementation and strategy.
And don’t miss the other episodes of App Marketers Unplugged for more insights!