A regional QA consultancy

How a QA consultancy moved mobile regression onto real phones

This consultancy runs regression testing for its clients' apps. It used to do that by hand. Now CodeReviewer writes the suites once, and they replay on the team's own Android phones while nobody's watching.

Drafted with AI assistance and reviewed by the CodeReviewer team. How we use AI

Real devicesevery replay runs on physical Android phones, not emulators
No rootlab phones paired by QR code, no rooting or laptop tether
Unattendedscheduled replays run without anyone at the bench
Evidence attachedevery failure comes with screenshots and a step-by-step trace
Industry
QA & testing services
Team
Small QA team
Region
Middle East & South Asia
Uses
Mobile testing, Test automation, Bug Explorer

The challenge

Most of this consultancy’s work is regression testing for client apps. Every new build means the same routine. Sign in, go through onboarding, place an order, check the receipt. Then do it again on the next phone.

They’d tried scripted automation before. It didn’t last. Every time a client shipped a redesign, the scripts broke, and fixing them took longer than just running the checks by hand.

Emulators had their own problem. The bugs clients reported from the field often never showed up on an emulator: a keyboard covering a button, a permission dialog only one manufacturer shows, a slow cold start on a mid-range phone. The team wanted to test on the phones real users actually carry.

The approach

They set up a small bench of Android phones and paired each one with CodeReviewer. Install the agent app, turn on accessibility, scan a QR code, and the phone shows up in the workspace as online. No rooting, no laptop plugged in.

For each client app, they upload the build and point CodeReviewer at the feature they want covered. It reads the relevant code, uses the app on a real phone the way a person would, and records a suite of test cases. Before anything gets authored, the team looks over the proposed test plan. They cut the cases they don’t need and add a few negative paths of their own.

After that, the suites replay on a schedule. When a client changes a screen, the broken test gets repaired and checked again instead of sitting red. Flaky cases are benched and retried on their own, so one shaky step doesn’t hide real failures. Between releases, the team points Bug Explorer at staging builds to find crashes and dead ends nobody wrote a test for.

The results

Regression passes that used to need someone’s full attention now run on the bench by themselves. Every replay keeps screenshots and a step-by-step trace, so when something fails, the reproduction is already there, ready to send to the client’s developers. The team spends its time on exploratory testing and talking to clients, not clicking through the same flows again.

Client name withheld at their request.

Give every PR the review it deserves.

Start your 15-day free trial. We onboard a few teams every week.