June 10, 2026 · The Caliper team

Apple said nothing about attribution at WWDC. That’s the news.

WWDC wrapped this week. Siri AI, iOS 27, a new Core AI framework — the keynote was an AI event top to bottom. If you work in mobile UA, the notable announcement is the one that didn’t happen: nothing on SKAdNetwork or AdAttributionKit. No postback changes, no conversion value updates, no crowd-anonymity adjustments, no re-engagement news.

That breaks a streak. Apple touched measurement at every WWDC in recent memory — AdAttributionKit was introduced in 2024 and got a substantial overhaul in 2025. This is the first quiet year since the post-ATT framework era began.

The two readings

The takes circulating this week split into two camps.

One: Apple has quietly concluded the framework isn’t worth further investment. Adoption has been grudging, the docs are famously hard to use, and most of the ecosystem treats SKAN numbers as one input to triangulate rather than a source of truth. On this reading, the silence is the sound of a project entering maintenance mode.

Two: Apple is working on something bigger and chose not to preview it. The speculation in this camp runs toward further restrictions on the signals that probabilistic attribution still leans on. Nobody outside Cupertino knows, and anyone who says otherwise is guessing.

It doesn’t matter which one is true

Here’s the thing — operationally, both readings say the same sentence: Apple’s measurement roadmap moves on Apple’s schedule, in Apple’s interest, and you don’t get a vote.

If reading one is right, the framework you’re building conversion-value schemas on is frozen, with all of its current limits — network-level resolution, anonymity thresholds, delays — now permanent features rather than v1 problems.

If reading two is right, some portion of today’s measurement stack gets invalidated on a date Apple picks. The teams hurt worst would be the ones leaning hardest on inference — probabilistic matching and anything else that reconstructs identity from signals Apple can coarsen at will.

Either way, the lesson isn’t “abandon SKAN.” It’s still the system of record for iOS paid UA, and you should keep using it for what it actually measures. The lesson is about concentration risk: if your entire understanding of where installs come from depends on one platform’s framework, your measurement strategy has a single point of failure that just demonstrated it can go silent for a year without explanation.

What doesn’t depend on Apple’s roadmap

Some measurement legs have no Apple dependency at all. Media mix modeling runs on spend and outcomes. Incrementality tests run on geography and budget. And self-reported attribution runs on asking the user — a first-party question, answered in your own app, attached to the install record you already own. No postback spec to chase, no anonymity tier to clear, no WWDC session to wait for.

We’d argue the quiet year is a good moment to audit the mix: which of your attribution signals exist at Apple’s pleasure, and which would survive any announcement next June? The first list shouldn’t be the whole list.

Time will tell what the silence meant. Your measurement stack shouldn’t have to care.