What you can still measure without the IDFA

What exactly is lost?

One thing: the device-level join. When a user does not give permission, the advertising identifier is all zeros and you may not track them.^[1] An attribution tool can no longer say "this click and this install were the same device" by comparing identifiers, which was a main form of deterministic matching on iOS.

Everything on the list below survives that loss.

What you can still measure

1. Installs per campaign

Apple's frameworks attribute the install on the device and report it to the ad network. Apps can call SKAdNetwork regardless of their tracking authorization status, and AdAttributionKit does not need the prompt either.^{[2, 3]} You get a count of installs for each campaign, without identities.

2. An early quality signal

The first postback can carry a fine-grained conversion value that you define.^[4] If you map it well, you can tell a campaign that brings trial starters from one that brings people who open the app once.

3. Later activity, in rough levels

SKAdNetwork 4 can send up to three postbacks across days 0 to 2, 3 to 7 and 8 to 35.^[4] The second and third carry a coarse value.^[4] When the second and third postbacks are sent, that is enough to see whether a campaign's users are still active or paying in week one and month one. At Tier 0 the system sends only the first.^[4]

4. Full detail for users who say yes

For users who allow tracking, the identifier is readable and device-level attribution is permitted.^[1] This group is a self-selected part of your users, so treat its patterns as indicative. It can show patterns that the campaign-level postbacks cannot.

5. Everything inside your own app

Apple's definition of tracking is about linking data from your app with data from other companies.^[1] Your own records of what signed-in users do in your product, such as subscriptions, retention and revenue, are still yours to analyse. They tell you how the whole user base behaves, even when they cannot tell you which ad brought each person.

What you cannot measure this way

The last row is the one that surprises small teams. See crowd anonymity for why low volume means less detail.

A reporting plan that fits these limits

1. Judge iOS campaigns on a weekly rhythm. Daily numbers are incomplete until the postbacks arrive.
2. Use the framework count as the campaign total and the consented group as a detailed, self-selected subset. Do not add the two together, because a consented user can appear in both.
3. Keep campaigns few and large enough. Splitting a small budget across many campaigns gives each one fewer installs, which can mean less detailed postbacks.
4. Compare against totals you own. Total new iOS users and total revenue from your own records, set against total spend, is a sanity check that needs no identifier.
5. Estimate long-term value from early signals. Since postbacks stop at day 35, long-run LTV by campaign has to be projected from what the early windows show.

For how the same questions are answered on the other platform, read how iOS and Android attribution differ.