SKAdNetwork and AdAttributionKit

Context only. This is genuinely an engineering and measurement topic; what actually matters here is understanding why paid numbers on this side of the centre behave the way they do.

What these systems actually are

One was this platform's first privacy-preserving way of crediting advertising, built on the anonymity of a large crowd rather than tracking any one shopper.

A newer system lets registered advertising networks and shop owners credit a walk-in or a return visit to a specific campaign, by receiving a signed signal from the centre itself that deliberately limits what data comes back, and it builds directly on the older system's own foundations.

For advertising running on genuinely outside platforms, the centre offers this newer system specifically.

How these differ from the centre's own first-party tool

The centre's own first-party tool is only for its own billboard, uses only the centre's own first-party data, and gives credit all the way down to the individual word. The newer signal-based systems are for any registered advertising network at all, use a signed signal from the centre rather than direct first-party data, deliberately limit their own detail, and genuinely can't see advertising outside the centre's own shopfront, which is precisely their whole purpose.

The centre describes its own first-party tool and the newer signal-based system as two freely available, complementary methods for its own billboard customers specifically.

Word-level credit exists only inside the centre's own billboard, because the same centre owns both the billboard and the shopfront itself. That's exactly why the whole word-harvesting loop is only really possible there, and nowhere else.

What these systems deliberately don't give you

Both of these frameworks limit what comes back by design. That's the whole point, not a flaw. In practice:

No detail on any one individual. Grouped together by construction.

Delayed. The signal arrives only after a deliberate delay, so reading a campaign's results on the very same day simply isn't available.

Coarse. Limited campaign labels and limited detail on what actually converted.

Subject to a minimum. A low-volume campaign can get less detail back, or none at all, the same broad kind of problem as a minimum reporting threshold anywhere else.

What someone doing this job actually needs from this

Mainly one thing: it explains why paid reporting on this side is slower, coarser and more argued-over than the other centre's. When a colleague working on paid growth says the numbers for this side aren't in yet, or that campaign-level detail is missing, this is why. It isn't a tool failing to work properly.

Second: it's exactly why the centre's own paid billboard is measured better than any other advertising on this side. The centre's own first-party tool gives word-level, first-party data that no outside network on this side can ever match. That real asymmetry is a genuine argument for the centre's own billboard, beyond its advertising value alone.

Third: none of this touches your own unpaid results at all. Your own back-office's own numbers are completely unaffected. Being found and chosen didn't get any harder to measure; only paid measurement did.

The rabbit hole

Encoding a conversion's value, signal tiers and their windows, the actual thresholds behind crowd anonymity, migrating between the older and newer systems. Real, deep, and owned by whoever actually owns measurement. If you're being asked to make a decision that depends on these specific details, the right move is simply to bring that owner into the room.

What tends to go wrong

Expecting same-day campaign data on this side. The signal is delayed by design.

Blaming your own scorekeeper for coarse data here. It's the underlying framework, not the tool.

Assuming the centre's own fine-grained detail applies to other networks too. Only its own billboard gets that.

Thinking any of this affects your own unpaid reporting. It genuinely doesn't.