In April 2021, iOS 14.5 shipped App Tracking Transparency, and user-level advertising identity became opt-in. Most users declined. With that, the industry's comfortable fiction — that every install could be deterministically traced to the ad that caused it — stopped working for the majority of iOS traffic.
What replaced it is honest about its limits: Apple's SKAdNetwork delivers aggregated, delayed, deliberately coarse signals — no user-level data, restricted conversion values, randomized timing. Android's Privacy Sandbox points the same direction. Ad platforms fill the gaps with modeled conversions: statistically estimated, not observed. The dashboards kept their decimal points; the decimals just mean less now.
The panic of 2021 obscured something important: the most decision-relevant numbers were never inside the attribution black box to begin with. Four things remained fully knowable:
| LAYER | WHAT IT ANSWERS | HOW |
|---|---|---|
| Spend-aware True ROAS | Is this campaign's spend working? | All-stream revenue of install cohorts against campaign spend — approximate, consistent, yours |
| Geo-level reads | Where is spend working? | Country P&L: revenue and spend per geo — precise without any device identity |
| Incrementality thinking | Did the spend cause anything? | When spend steps up or pauses, does portfolio net move? Budget changes as experiments |
| Time cohorts | Is quality holding? | Revenue curves by install week — payback windows, decay, retention economics |
Cohort attribution of ad revenue to campaigns is an approximation — a user acquired by Campaign A and monetized for months isn't tagged with A in your ad network. Approximate and consistent beats precise and imaginary: the same method applied every week ranks campaigns correctly even when absolute values are soft. That ranking is what verdicts need.
Platform-reported ROAS didn't become useless; it became directional. Three rules keep it useful:
The uncomfortable truth of the privacy era is also its gift: it forced measurement back onto the numbers that were always the most honest — what you earned, what you spent, where, and when. Publishers who rebuilt on those four are measuring better today than the industry ever did in the deterministic years. The ones still chasing user-level certainty are optimizing a memory.
Connect your first platform in minutes and see what your portfolio actually earns — after spend, every day.