SKAdNetwork 4.0 officially launches for iOS and iPadOS 16.1 (and yes, there are changes)

Apple promised SKAdNetwork 4 before the end of the year. Apple delivered.

To me, when Apple promised SKAN v4 in 2022, that meant likely November, but potentially December. Perhaps December 29th, or maybe even the 31st. However, SKAdNetwork 4 is now officially live and released, and it’s only October 24. In fact, not only is it announced, SKAdNetwork 4.0 is actually in the wild right now, included in iOS and iPadOS 16.1.

The newly updated documentation highlights some important changes and confirms that web-to-app SKAdNetwork postbacks are Safari only, not Chrome or other browsers. Briefly, here are some of the bigger reveals in the updated SKAN documentation:

• Lockable postback windows
• Longer random timers for postbacks 2 and 3
• More details on crowd anonymity

Lockable postback windows for faster conversion value delivery (sometimes Apple gives)

If there are sufficient app installs in a campaign to ensure crowd anonymity, an app install can get up to 3 individual SKAdNetwork 4 postbacks after 3 separate periods of time, which we already knew. (Note: they are not connected, so you will not have the ability to tie postback 2 and 3 for a specific app install to the first postback for the same install.)

Here are the postback periods:

The new information just released today is that instead of continuing to update the conversion value that an SKAdNetwork 4 postback will carry for the full period of time (first: 0-2 days; second: 3-7 days; third: 8-35 days) the advertised and now-installed app can instead choose to “lock and finalize” a conversion value earlier. 

For example: someone makes a purchase in your retail app on D0 or D1 (essentially immediately). You can lock the conversion value then, and kick off a postback immediately:

In the above example from Apple, the advertised app successfully generated a conversion event that the marketing and product teams view as significant enough to want to know about ASAP. This even happens at some point before the end of the second postback period (in the image above, it looks like D5 or so). 

Via predetermined logic, the app developer locks in and finalizes the value, which prompts SKAdNetwork 4 to immediately prepare a postback and ignore all further input during that conversion window. 

You don’t get the postback immediately, however. Locking the postback window instead kicks off the random time earlier, and you’ll actually get the postback within 24 to 144 hours (second and third postback; earlier for the first). So you get it earlier than you otherwise would have … but you don’t get it immediately.

On the one hand, this is good news: you can get data quicker (though there’s a caveat, see below).

But it’s also potentially challenging, because different ad partners may — as they have in SKAN 3 — want different things. For instance, many may default to the earliest possible conversion windows in order to get data for optimizing campaigns. In this scenario, ad networks may ask advertisers to lock SKAdNetwork 4 conversions after D1 to get postbacks within 48 to 72 hours after install, versus 72 to 96 hours.

Longer random timers for postbacks 2 and 3 in SKAdNetwork 4 (sometimes Apple takes away)

Getting postbacks earlier by being able to lock them is good.

But … there’s a catch.

Way back in SKAN v2, Apple implemented a 24-48 hour timer during which, randomly, SKAdNetwork sends a postback. In SKAN v4, Apple kept that exact timer for the first postback, but increased the length of the timer for postbacks 2 and 3.

Now, instead of getting a postback somewhere between 1-2 days after the conversion value is finalized, you’ll get a postback 1 to 6 days after the conversion value has been finalized. You read that right … up to 6 full days until you’re certain to get postback #2 and postback #3.

There’s a couple things going on here that aren’t ideal:

1. Later data is just generally bad for marketers
2. Delayed data on these postbacks makes it much harder to understand performance

You’re essentially getting a postback potentially up to 6 days after a conversion value event. Remember, conversion value events are significant times, important things happening in your app that you’d ideally like to be able to use to understand the value of ad campaigns, partners, and channels. Now the data you need will be significantly delayed and much harder to use. For example, if you want 7-day revenue, you’ll essentially get it potentially as late as 2 weeks after the install.

One speculation for the cause: Apple might want to shift daily granularity to weekly. Another: additional delay simple adds to anonymity.

More details on SKAdNetwork 4’s crowd anonymity

Apple says that it will take 3 factors into account when determining the crowd size for its privacy-ensuring crowd anonymity calculation, which happens at the point of actual install:

1. Crowd size for the app or domain displaying the ad
2. The advertised app
3. The hierarchical source identifier from the ad network

Assigning a crowd anonymity score seems to be based on the last click and assigned at time of app install. 

Where SKAN v3 only enabled 2 digits of campaign IDs so you only had 100 possible campaigns (0 to 99), SKAdNetwork 4 allows the first postback to have up to 4 digits in what is now called Source ID.

At the lowest levels of crowd anonymity, where Apple must censor data in order to preserve privacy, you’ll still only get those last 2 digits that basically correspond to the 2 you got from SKAN v3. With higher levels of installs per partner, however, you open up the possibility of getting additional digits to unlock extra layers of reporting granularity.

Apple pictures the tiers this way:

SKAdNetwork will compute the data tier, and, if there are enough installs to warrant it, returns a full data payload in the first postback — the only one that can have a fine conversion value with more data. That will enable you to receive a very small amount of extra encoded data for whatever you choose: creative optimization, location, sub-campaigns, ad placements, ad type, whatever you wish that fits the parameters …

Expect adoption to take some time

Getting SKAN v4 now is a win. But don’t expect to be seeing this used in volume immediately. Marketers will only receive SKAdNetwork v4 postbacks if each of the 3 following conditions are met, Apple says:

• The ad network showing the ad generates v4 ad signature
• The app showing the ad is built with iOS 16.1 SDK or later, or, for web ads, the browser is Safari 16.1 or later
• The app being advertised is App Store-signed and is running on a device with iOS 16.1 or later

iOS users update their operating systems pretty quickly, so that’s not a primary concern. But app developers take some time to update SDKs and apps, and ad networks also need some time to include SKAdNetwork 4 in their plans as well.