VIEW THE SLIDES: SLIDES FROM PRESENTATION HERE



ABOUT ROCKETSHIP HQ: Website | LinkedIn  | Twitter | YouTube

KEY HIGHLIGHTS

⚙️How does SKAdNetwork work?

⚖️How is conversion tracking different with SKAdNetwork vs. MMPs?

💰How might you define ‘conversion value’ in SKAdNetwork?

🔺How does targeting change under SKAdNetwork(vs. the current paradigm)?

🏗How does campaign structure change with SKAdNetwork(vs. the current paradigm)?

🤑How do you measure and report on performance under SKAdNetwork?

FULL TRANSCRIPT BELOW

Hey everyone –

Welcome to today’s session. I’m presenting a playbook for marketers to succeed in a post-IDFA world. 

I’m assuming you have some background about IDFA deprecation and how it’s impacting the ecosystem. If you are looking for basics, please check out our 101 webinar on Youtube, iTunes or MobileUserAcquisitionShow.com – it’s titled RIP IDFA: How to navigate a post-IDFA world with iOS 14.

Today we’ll talk about what is changing for marketers – and how your playbook as a marketer is expected to change with iOS 14 and SKAdNetwork.

Huge thanks to a lot of people whose important notes on a lot of forums forms a lot of this presentation of my thinking — Gadi from Singular, Offer from Fyber, Maor – Ex-AppLift, Eric Seufert from Mobile Dev Memo, Thomas who’s an independent consultant, Nebo from N3twork, Pau from Goodgame, and Eran from Singular.

But here’s a bit about me — I’ve been doing UA since to 2013, it seems like a very different time. Please check out mobileuseracquisitionshow.com and rocketshiphq.com for more stuff like this. 

**

May you live in interesting times. 

This was a (possibly apocryphal) curse – and I like to include this to highlight that these are times of opportunity and challenge.

**

This is what we’re going to cover today. 

What’s SKAdNetwork? How’s conversion tracking different? — which is, I think a very good portion of what is going to change come September. A lot of user acquisition, a lot of processes, a lot of our workflow is going to change. How’s the targeting changing? How’s our campaign structure changing? How’s our measurement and reporting changing? What is SKAdNetwork? How is it different from our historical MMP? 

**

Fundamentally here’s what’s changing:

Historically we knew where each individual user was coming from – and how valuable each user was.

Now we know where each ‘set’ or ‘campaign’ of users is coming from – and it’s going to be extremely challenging to know what their exact value is.

**

Here’s how this looks in practice.

In the MMP paradigm, you know precisely the value down to the user as to what actions they performed.

Now under SKAdNetwork you have a ‘conversion value’ – which may or may not correspond to actual revenue of each user. And you have this at the campaign level, not at the user level.

**

To look at this in more detail, with an MMP you clearly understand every event that a user performs – IAP 1, IAP 2, level 20, level 40 – what have you. With all of this data, you can essentially calculate the exact LTV of a user.

With SKAdNetwork, you cannot do this – because all you have is a conversion value – which *could* be extrapolated into an LTV – but that is hardly going to be accurate.

On the next slide, I’ll talk about some of the challenges with the conversion value – why it doesnt correspond to actual revenues – and what to do about it.

**

Why is conversion value imprecise under SKAdNetwork?

Let’s look at a user’s flow within the SKAdNetwork framework.

A user installs – and a postback is sent to the ad network: hey you sent this install, and their value is zero.

A user makes a first purchase – Apple says: hey now the user’s value is $1

A second purchase -> Apple says: hey now the user’s value is $2.

Now: if a user keeps making transactions every 24 hours, amazing -> you can record up to 64 events – and you have a clear read of how much revenue this user has made in their first 64 transactions.

The reality is that most users wont make transactions every 24 hours. In fact, for many apps there wont be a transaction in the first 24 hours – and their value will get recorded as zero.

You might have very valuable users – and have no way to measure the value of campaigns that sent these users if they make a massive transaction 48 hours after installing.

**

The way I recommend thinking about this is to define values for non-purchase events – and set these up in a way that highlights who is a valuable user – and who isnt – and what that continuum looks like.

To look at a concrete example….

Let’s say a user installs, registers after 22 hours, goes to level 1 after 36 hours, goes to level 50 after 50 hours – and makes their first purchase after 60 hours.

This other user installs and registers – and never returns – and is a much less valuable user.

By assigning non-monetary value to these events, you can arrive at a ‘score’ or conversion value – even though the purchase does not happen in the first 24 hours.

What I recommend doing is looking at your current sequence of events to ascertain what events and values make sense to use in your conversion calculations.

**

And if you want inspiration for ways to think about it, our friends at Singular put out a spec about how they are thinking about using SKAdNetwork – which has some useful pointers.

You can set up events based on revenue – assuming you have a transaction occurring within the first 24 hours(which for many apps can be rare). 

For some subscription based apps, there’s a free trial within 24 hours – and those are the easiest to model. For high-ARPU apps that have a transaction further downfunnel, this becomes much much more challenging.

You can set up events based on retention – again this makes a lot of sense for many apps since it just requires a user to open the app, they dont have to do anything.

You can of course set up a funnel based optimization – wherein you set up values based on steps in a funnel.

Eventually what matters is that you are careful and intentional about the values you pick. And your existing user flow is the best place to start looking at as a starting point for understanding what system of values makes the most sense.

**

Let’s look at targeting, campaign structure and measurement next.

With targeting, on Facebook, you use first party data(email IDs or past IDFAs) to make lookalikes; or use purchasers tracked by Facebook SDK to do your targeting.

On all channels, you could basically run broad campaigns – much like UAC. Or use interest & demo based targeting – as is fairly common with other SANs.

You can use creatives for defining targeting – say show older people in ads to target older demos, show certain types of puzzles to target users who play certain types of puzzles – and so on. 

**

Here’s how the campaign structure changes. 

You have up to 100 campaigns – and sources within each campaign. Each campaign can be defined for a geo, placement or audience.

What is still TBD is whether sources can correspond to ad sets or ads – or if it’s just source apps by themselves. Apple’s documents say that it has to be an iTunes application – but it remains to be seen as to how Facebook and Google will plug into SKAdNetwork.

**

With measurement, we see spend, installs and conversion value for each channel, campaign and source – as reported to each ad network.

You as an advertiser still need to aggregate these and put these together – and this is a role that MMPs might end up playing in the future.

** 

Q&A

Q: Facebook SDK, firebase SDK, hashtag collect data from almost every app in the world. IF most of the signal will be gone, how do you think it will impact the performance?

A: I wouldnt say that most of the signals will now be gone. Certainly an important part of the signal will be gone. Facebook has IDFVs of users on the Facebook family of apps. Facebook has purchaser behavior that is associated with these IDFVs. Facebook knows and understands this performance. So I daresay Facebook will build an internal device graph. That is my strongest hypothesis to the question. 

So, I would say short term, I think there will be a dip because the strongest signal is going to go away. Longer term, I think they’re going to adapt by building a new device graph based on their own owned and operated first party data.

Q: How does the new Apple privacy messaging allow publishers to use historical data for ongoing targeting?

A: I don’t think it forbids using historical data — please correct me if I’m wrong. I have not seen anything in the documentation that forbids using historical data. It says future data — “hey, would you like to be tracked in the future? Yes or No.” That is the spirit of the prompt. And certainly if you have emails, I’m imagining that it is collected with some solid consent. So email first party data can be captured.

Q: Do you think Facebook will offer some tools to map SKAdNetwork to the SDK?

A: Unofficially, I have heard yes. Unofficially, I know that they are working on this. This is a very high priority for them. They are working so as to be plugged into SKAdNetwork just as well. 

Q: For small developers, this SKAdNetwork implementation seemed complex. Is there a way to simplify it?

A: I would say yes, there is a way to simplify it. I have a number of subscription apps for which I think the SKAdNetwor implementation is basically going to be a free trial within 24 hours — it’s going to be a blessing for them. So yes, it can be more complex. And I think the screenshot I showed for Singular’s documentation, I think that shows the kind of complexity that is possible. So my encouragement really would be to start by just keeping it simple — register the day zero or day 1 retention, first time user experience onboarding, first user purchase. Hopefully that works.

I think it can be a little more challenging with high ARPU apps. So I do agree with that. For high ARPU apps, I think they’re going to have to find a way to simplify their event structure, maybe put a starter pack within the first 24 hours. 

Q: The concern is not that the conversion value is not encrypted, but it’s not signed.

A: Thank you for correcting me, which means there’s no way to authenticate that the conversion value was not changed by the network. So basically, the conversion value can be manipulated, which is kind of crazy that Apple didn’t contemplate that. 

Q: How will DSPs and retargeting companies be impacted without a device graph?

A: Yes, so DSPs and retargeting companies will need to be on the P list of publishers to get a postback. Any entity that requires that their traffic be tracked needs to be on the P list of publishers. So yes, how they will be impacted without a device graph, I think we had an episode of Mobile User Acquisition Show two days ago, which goes into this in a lot more detail — Pau Quevedo’s whose inputs also inform this.

Q: Will Apple’s revenues be impacted by this?

A: Downstream I think. This is something we can conjecture. But I would say this downstream effect — if people advertise less on Facebook, app developers get fewer revenues. The total downloads of apps drop, then downstream there could be an effect. In the comments, I see that Apple has been building an army of media CEOs client relations, which may be true, but I think Apple does not have a good track record of building ad tech. I think I mentioned it the last time. So yes, Apple may be building an army of media sales and client relations, but I don’t know if I would bet on them building the sort of ad tech that can really capture any of the dollars they may be displacing to this. 

I think short term there could be an impact. Longer term, I would say within three to six months, people need to grow, the apps aren’t going away. I think people will work with the structure — the sort of playbook I described. People are working with Google UAC today. If you’re working on iOS, you don’t get search traffic. On Google UAC search, you’re basically operating as if there’s just no idea where your traffic is coming from. I would say people that get accustomed if I were to conjecture, and this is a conjecture, I think people will get used to more of a UAC-like setting, and it causes closer to a medium mixed modeling methodology where there’s more reliance on incrementality. I think over the next three to six months, things will even out. I think that’d be a shotgun dip, then people start to figure this out. 

**

Overall: expect to change all parts of how you run your marketing: strategy, targeting, campaign structure and reporting.

I’d recommend starting to prepare today. Please hit me up if you have questions. Check out MobileUserAcquisitionShow.com – where we’ll have a lot more about this & other things user acquisition.

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 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!