Documentation on phased releases via App Store Connect is hard to come by. The basics are easy to grasp, but some details around the edges are a bit fuzzier. How does it actually behave if you pause, restart, or issue a new release during a phased release?

My team and I recently released a major update to an app with 3.5 million users, giving us the opportunity to see how phased releases work when you push them. It’s important to note that the detail here is based on observation, not documentation (which seems to be impossible to find).

When talking about percentages in phased releases, I’ll use users as shorthand to mean users with auto-update switched on. And conversely when I talk about releases going out to 100%, that only applies to users with auto-update switched on. You can’t guarantee that users will ever update an app version.

• Phased releases steadily increase the number of users who get the update. The percentage of users increases every 24 hours as follows: 1%, 2%, 5%, 10%, 20%, 50%, 100%. I.e., it roughly doubles every day until all users have it.
• You can pause and unpause the release at any time, but the total amount of time it can be paused for is 30 days. I haven’t tested what happens if you reach 30 days whilst paused, but the implication is that it will go out to 100%.
• You can pause a release, but you can’t stop it or roll it back. The only way to stop it going out to 100% is to release a version that supersedes it.
• When you pause a release, it stops the percentage ticking up to the next level. It doesn’t stop new updates. So if you’re at 1%, and you pause, you’ll see the number of users on the new version continue to update until 1% have it installed. If you’re planning to pause, it’s important you do it before it hits the next 24-hour mark. Otherwise the number of users will (approx.) double again, and you won’t be able to stop it.
• The number of users that receive updates at each stage was higher than we expected. The estimates we had were based on analytics showing the number of active users. The release percentages are based on installed users, which turned out to be around 10 times higher than we were expecting.
• Regardless of what percentage you’re at on the phased release, all new downloads will be the new version, and any user who chooses to manually update will get the new version. Be cautious of any advertising pushes that could ramp up your numbers when you don’t really want it.
• Users are randomly selected at each stage of the phased release. If you find a bug during the rollout, and you start the phased release again with the fix, new users will be randomly selected again from the whole population. Users that got the version with the bug won’t get any preferential treatment, and they could be in the last tranche of users to get the updated version.