Mobile app accessibility testing with VoiceOver and TalkBack
Start mobile app accessibility testing by completing a real task with VoiceOver on iPhone and TalkBack on Android. Check whether you can reach each control, understand its spoken description and finish the task. When something fails, record the gestures, focus sequence and words you heard alongside the build and device settings so a developer can repeat it.
What should you prepare before testing with a screen reader?
Choose a journey with a clear result: sign in, change a delivery address or submit a support request. Write down the starting screen, test data and expected outcome. Test the whole journey, including errors and confirmation, rather than stopping when every button has a label.
Use the build you intend to release and a test account without real customer information. Record the phone model, OS version, app version and build number. On Android, include the TalkBack version; also note the speech language and any changed verbosity, gesture or reading settings. Google warns that TalkBack behaviour varies by device, Android version and TalkBack version. Use the mobile device test matrix to keep the tested combinations and gaps visible.
If you're new to the gestures, practise before testing the app. On iPhone, open Settings → Accessibility → VoiceOver and use the practice area. Apple explains how to turn on and practise VoiceOver. For Android, use the TalkBack tutorial on the test phone. Learn how to turn the service off again before starting the run.
How do you test a journey with VoiceOver or TalkBack?
Move through the screen in sequence using the screen reader's navigation gestures, then explore by touch. A control you can find by touching its known position may still be missing from the sequence someone uses to discover the screen. Use the device's tutorial if gestures have been customised; Google's TalkBack testing guide covers sequential navigation and exploring by touch.
Keep the screen reader on while entering data, opening menus and confirming the result. For each case below, record Passed, Failed, Blocked or Not run against the tested build.
| Check | What to do | What to listen for or observe |
|---|---|---|
| Find the next action | Navigate from the screen's starting focus through its content. | The controls needed for the task are reachable in an order that makes sense. |
| Understand a control | Focus an icon button, text field and selectable option. | The description explains the purpose; relevant roles, values and selected states are understandable. |
| Activate and return | Open a menu or dialog, use it, then dismiss it. | You can operate the new content and find your place again afterwards. Check that background controls don't interrupt a modal task. |
| Correct an error | Submit missing or invalid test data, then fix it. | You can discover the error, identify the affected field and reach it without losing the other entries. |
| Notice a change | Trigger loading, a saved result or another status update. | You can tell what happened and what to do next without relying on a visual change alone. |
| Use a long list | Navigate beyond the initially visible rows, then open and return from an item. | More items remain reachable, row descriptions distinguish them and returning doesn't make the task confusing. |
| Use a gesture action | Try a task normally completed by swiping or dragging. | A screen-reader action or another accessible control lets you achieve the same result. |
Don't prescribe an identical spoken sentence across iOS and Android. Check that the output gives the information needed for the task, then quote the actual output in a failure report. A quantity control that only says “button” leaves a different problem from one that names the item but fails to announce the changed quantity.
When the app opens a web view or hands a task to a system screen, continue testing through the transition. Record where a failure occurs. “Checkout inaccessible” doesn't tell engineering whether the problem is in the app's basket, the payment page or the return to the app.
Which checks belong alongside the screen-reader pass?
Repeat the journey with larger system text and display settings. Look for clipped instructions, overlapping fields and actions pushed off screen. On iOS, include the larger accessibility text sizes. Apple's guide to testing system accessibility features covers Dynamic Type, contrast, reduced motion and keyboard interaction. Record the settings used so a layout failure can be reproduced.
Check whether errors, selection and progress remain understandable without colour alone. Plan separate cases for keyboard, switch or voice input where those methods are relevant to the app's users, and captions for audio content. List checks you haven't run in the test summary report. Record which journey passed, on which device and with which settings.
Include people who use assistive technology in usability testing. They can reveal difficulties that a tester learning the gestures may miss. Android's accessibility testing guidance includes user testing alongside manual, automated and tool-based checks.
What can automated accessibility tools tell you?
Use automated checks to find issues worth investigating, then repeat the affected task manually. Android's Accessibility Scanner can suggest improvements to labels, touch targets and contrast. A useful label still needs to make sense in context: “Remove” on each basket row may pass a basic label check while leaving the shopper unsure which product they will remove.
On Apple platforms, Accessibility Inspector audits help identify potential issues. Keep the audited screen and settings with each finding. A clean audit of one screen doesn't tell you whether someone can recover from a sign-in error or finish a payment. Attach the scanner result to the ticket and explain which task the issue prevents.
How do you report a screen-reader failure so someone can reproduce it?
Name the task that is blocked and the condition that triggers it. Here's a fictional example:
TalkBack skips the delivery option after an address error. On the recorded Android build, enter an invalid postcode and submit. Correct the postcode, then navigate forward from the postcode field. Focus moves to “Place order”, skipping the delivery option. Exploring the delivery row by touch also finds no selectable control. The shopper cannot choose delivery without turning TalkBack off.
The report should then give the exact build, phone, settings and steps. If you don't know why the control disappeared from navigation, leave the cause open. Engineering can inspect the accessibility information after reproducing the failure.
Copy these fields into the issue:
Title: [screen reader] + [failed action] + [trigger]
Task blocked and user impact:
App version and build:
Device model and OS:
Screen reader and version where available:
Speech language and relevant settings:
Text size / display size / orientation:
Test environment and account alias:
Starting screen, data and initial focus:
Steps, including gestures or screen-reader actions:
Focus sequence around the failure:
Actual spoken output (or no announcement observed):
Expected result:
Actual result and any workaround:
Failures / attempts with this setup:
Sanitised evidence and timestamps:
What the capture does not show or record:
Other tested configurations and results:
A screenshot can show the screen and sometimes the focus indicator, but it doesn't preserve the spoken feedback. If you attach a recording, play it back first and check whether it actually includes the screen reader's speech. Write the relevant words and focus sequence in the ticket even when you have a video. Remove private information from both the picture and the audio.
After a fix, rerun the original steps with the same settings, then complete the surrounding journey. For a focus fix, check the controls before and after the changed element as well as the element itself. Keep the original failure beside the retest result; the mobile bug retesting checklist covers that handoff.
Use the bug report template generator to format the failed case for your tracker, adding the screen-reader settings and spoken output to its reproduction steps and actual result.
Frequently asked questions
How do you start accessibility testing for a mobile app?
Choose a real task, record the app build and device settings, then try to complete it with VoiceOver on iPhone or TalkBack on Android. Check whether you can find, understand and activate each control. Add separate checks for larger text, contrast and other input methods; one screen-reader pass does not cover all accessibility needs.
Can automated checks replace VoiceOver or TalkBack testing?
No. Automated checks can flag issues such as missing labels or small touch targets. You still need to try the journey with the screen reader to judge whether the labels, navigation and changing content let someone complete the task. Include people who use assistive technology in usability testing.
What should an accessibility bug report include?
Include the blocked task, app build, device and OS, screen reader and relevant settings, starting screen, gestures, focus sequence, actual spoken output and expected result. Add a sanitised screenshot or recording if it helps, and write down anything the capture could not show or record.