What Should You Do When an App Update Breaks a Feature?
When an app update breaks functionality, first stop more users from being affected and protect any work already in progress. Then choose the fastest recovery you can safely verify: disable the affected feature, make a server-side change or publish a corrective update. Treat users who have already installed the broken version as a separate recovery problem; do not promise that pausing distribution will restore their previous app.
What should you check before choosing a fix?
Start with the user action that fails. “The app is broken” is too broad to guide a recovery decision.
Establish:
- The affected action: Can users open the app, sign in, save work or complete the main task?
- The affected versions: Does the problem happen after upgrading, after a fresh install or in both cases?
- The consequences: Is the action unavailable, or could it produce a wrong result?
- The scope: Does it affect everyone, particular devices or accounts with certain data?
- The current state: Was the user's action rejected, completed or left uncertain?
Prioritise possible data loss, incorrect transactions and uncertain outcomes over visual defects. If an action appears to fail, check whether it actually completed before recommending another attempt.
Capture enough evidence to reproduce the problem, without asking users to send passwords or unnecessary personal information.
When should you pause distribution?
If the update breaks a core action or could cause harmful results, pause further distribution wherever the release controls allow it. Check which release is active and whether its rollout is still limited before changing anything.
A pause is a containment decision. Plan separately for people who already have the affected version.
On Google Play, halting a staged rollout stops the release from reaching more users, while users who already received that version stay on it. See the official staged rollout guidance.
Do not assume that removing the release from distribution, republishing older code or changing a store setting will downgrade installed apps. Verify the available recovery route before communicating it.
For future launches, include the available pause controls and recovery steps in your Google Play release checklist.
Should you disable the affected feature?
Disable a feature when leaving it available creates more risk than temporarily removing access.
For example, if an updated booking form shows an uncertain result after submission, consider suspending new submissions until existing bookings can be checked. Leave unaffected actions available where practical, such as viewing previous bookings.
A remote feature control is useful only if the installed app already supports it. Before relying on it, verify that affected versions receive the setting and handle the disabled state correctly.
Check the whole journey:
- Does the app explain why the feature is unavailable?
- Can users still reach it through another screen or a saved link?
- What happens to work already in progress?
- Is there a safe alternative?
- What happens if the phone is offline?
Disabling an entry button is insufficient if another route still permits the harmful action. Verify the restriction wherever the action can be completed.
When is a server-side change the fastest safe recovery?
Consider a server-side change when the failure comes from something your service controls, such as a changed response, a validation rule or an unavailable service.
Suppose an update expects a response value that the service no longer provides. Restoring a compatible response may be a recovery option if testing confirms that it works for affected and older app versions.
Treat that as a diagnosis to prove, rather than a reason to change the service immediately. Confirm that the proposed change:
- Restores the failing user action.
- Preserves access and validation rules.
- Works with app versions still in use.
- Handles existing records correctly.
- Does not repeat an action already completed.
Do not weaken account protection or accept invalid data merely to make an error disappear. If the defect sits in code running on the phone, a service change may provide only partial relief—or none.
When do you need a corrective app release?
Prepare a corrective release when recovery requires changing the installed app. Examples include a broken screen, an incorrect calculation performed on the phone or a failure while updating locally stored data.
Keep the change focused on restoring correct behaviour. Postpone unrelated improvements so the recovery is easier to assess.
Test the journey users actually face: upgrading from the affected version. Also test an upgrade from the last working version and a fresh installation. These are separate acceptance checks; passing one does not establish the others.
Include interrupted actions, saved drafts, offline use and reconnection where relevant. If records may already be wrong, distinguish between preventing further errors and repairing earlier results. A corrected screen alone does not prove that existing records are correct.
Our guide to testing an Android app before release can help structure this verification.
How should you choose between the recovery options?
Use the least disruptive option that restores a safe, verified user journey. Some incidents need several measures in sequence.
| Situation | First action to consider | What must be verified |
|---|---|---|
| The update breaks a core journey and is still being distributed | Pause further distribution | Which users already have it and how they will recover |
| One feature produces unsafe or uncertain results | Disable the affected action | Alternate entry points and work in progress |
| A service change broke compatibility | Restore compatible service behaviour | Affected and older app versions |
| The fault is in the installed app | Prepare a corrective release | Upgrades, fresh installs and affected data |
| Earlier actions may have produced incorrect records | Contain the action and assess affected records | Repair accuracy and duplicate prevention |
For an illustrative booking example, the sequence might be: pause distribution, suspend new bookings, check uncertain submissions, publish a verified correction and then reopen booking.
Success means users can complete the task correctly and understand what happened to earlier attempts. It is more than getting the screen to load again.
What should you tell affected users?
Explain the impact, the safe next action and what recovery is available. Avoid technical explanations that do not help someone decide what to do.
A notice for the illustrative booking problem could read:
> Some booking submissions are showing an error. If you already submitted a booking, check your booking history before trying again. New bookings are temporarily unavailable while we check the issue. Existing bookings remain available to view.
Use wording like this only when each statement has been verified. If booking history is unreliable too, do not direct users there.
Keep the message consistent across the affected screen, help page and support replies. Say whether users need to install an update or take another action. Give a recovery estimate only when there is a sound basis for it.
Avoid instructions to uninstall, clear app storage or repeat a transaction until their consequences have been checked for the affected app.
What common mistakes make recovery harder?
Treating a distribution pause as a complete fix. Keep a separate recovery checklist for people already using the affected release.
Choosing a fix before reproducing the failure. An apparently similar error may have a different cause. Establish the failing action and expected result first.
Restoring access before checking existing work. Review uncertain submissions and damaged records before inviting users to continue.
Testing only a fresh installation. Verify upgrades with representative saved data.
Bundling recovery with new features. Keep the corrective release focused.
Announcing success after one successful attempt. Check the affected conditions and confirm that users have a practical route to the correction.
A short recovery review should feed into the next release checklist and your ongoing app maintenance plan.
What are the frequently asked questions?
Can we roll everyone back to the previous version?
Do not build the recovery plan around that assumption. Confirm the platform's supported options and how they apply to installed copies. Prepare a route for affected users that does not depend on a downgrade.
Should we always take the whole app offline?
Assess the consequences first. If one action can be safely isolated, consider disabling that action while keeping useful, unaffected features available. Restrict the whole app when narrower containment cannot protect users.
What if there is no remote feature control?
Assess whether the service can safely reject the affected action and return a message the installed app handles correctly. If that is unavailable, focus on distribution controls, clear user instructions and a corrective release.
When is the incident resolved?
Close it when the affected journey works, earlier uncertain results have been addressed, and users have clear recovery instructions. Record the failed scenario as a repeatable release check so the next update is tested against it.