lockWindow is a feature in Apple’s SKAdNetwork version 4 that allows app developers and advertisers to end a measurement window early. When a window is locked, Apple finalizes the conversion value and prepares the postback, instead of waiting for the full default measurement period to run.
SKAN 4 introduced three conversion windows, which together cover up to thirty five days after a user first opens an app. With lockWindow, teams can choose to finalize results earlier within any of these windows, allowing postbacks to arrive sooner and accelerating optimization cycles.
SKAN 4 tracks post install behavior in three windows:
Apps can update the conversion value inside each window. When a window ends, the value freezes. Apple then adds a random delay before sending a postback. The first postback has a shorter delay range than the second and third.
Without lockWindow, each window runs its full duration. With lockWindow, you can choose to stop the window early and lock the value as soon as you believe you have collected the key data for that stage.
Think of lockWindow as a manual close button for a measurement period. When you trigger it:
The earlier you lock, the faster you receive the postback. However, any events that happen after locking inside that same window will be ignored.
The random delay still applies. lockWindow does not guarantee instant delivery, it simply starts the delay sooner.
SKAN postbacks arrive with privacy based delays. In fast moving performance environments, early insight can help you:
At the same time, locking early means collecting less behavioral data. It is a trade off between speed and depth.
Use lockWindow when most valuable early events happen quickly after install. For example:
Most teams consider using lockWindow only in the first conversion window. Later windows often contain slower signals like subscription decisions or return engagement patterns, and cutting them short risks losing important insight.
Avoid locking too early when:
If you are unsure, start without locking and observe how quickly meaningful events cluster. Adjust with evidence.
The goal is to lock only after enough data has been gathered to make reliable performance calls.
No. It finalizes the value and starts the random delay earlier, but the delay still applies.
Yes. You may lock each window independently.
They are ignored for that window. Future windows still run normally unless you lock them too.
No. It is optional. It is a tool for faster feedback when appropriate.
Not always. Only do so if your most meaningful conversion events reliably occur early and your team benefits from faster feedback.
lockWindow does not bypass SKAN privacy rules. Crowd anonymity thresholds and coarse or fine value eligibility still apply.