SKAdNetwork postbacks and their three windows

What is a SKAdNetwork postback?

It is the message that tells an ad network one of its ads led to an install, without saying who installed. Apple describes SKAdNetwork as an API that helps advertisers measure the success of ad campaigns while maintaining user privacy.^[1] The postback is how that measurement leaves the device. It can carry a conversion value, and it does not tell you which user installed.

Your attribution tool collects these postbacks from the ad networks and builds the iOS campaign report from them.

What are the three conversion windows?

SKAdNetwork supports three conversion windows, which may result in up to three postbacks for a winning ad attribution.^[2] Each window is a period in which your app can update what it reports about the user.

The difference between the two kinds of value is covered in fine and coarse conversion values.

When do you get all three?

Two conditions come from Apple's documentation.

1. The ad is signed with version 4. The ad network needs to sign the ad using SKAdNetwork 4 or later.^[2]
2. The device runs a recent enough system. Apple says that starting in iOS 16.1, apps can update conversion values during three conversion windows, which results in up to three postbacks for an ad signed using version 4.^[1]

A third factor is outside your control. To maintain privacy, the device may limit the data in postbacks, and Apple assigns a postback data tier to each app download.^[2] If the tier is Tier 0, the system sends only the first postback.^[2] The idea behind the tiers is explained under crowd anonymity.

Why are the numbers always late?

The delay is deliberate. The usual explanation is that a postback arriving the moment a user did something could be tied back to that user by its timing. So the window has to close first, and then the random delay runs: 24 to 48 hours for the first postback, 24 to 144 hours for the second and third.^[2]

Say a fitness app is installed and first opened on a Monday. The conversion window begins at that first launch.^[2] The first window covers days 0 to 2 from that first open, and the first postback follows one to two days after the window. The team sees the install in its SKAdNetwork report late in the week. The third postback for the same install reports on days 8 to 35 and arrives more than a month after the install.

This changes how you run a test. A new iOS creative cannot be judged the day after launch from SKAdNetwork data alone.

What to do with this

• Leave an iOS test running long enough for the first postbacks to arrive before you compare creatives.
• Put the event you care about most where the first postback can report it, because only that one can carry the fine-grained value.^[2]
• Ask your provider whether its report shows the second and third postbacks, and how it labels installs that only produced one.
• Check how the newer framework changes the picture in AdAttributionKit and how it relates to SKAdNetwork.