Test strategy and plans
A written plan that lists the parts of the product, ranks each by what a failure would cost, names the method for each and sets the conditions for calling a release ready. You receive: the test plan and the case list.
QA and software testing
We write the test plan, run manual and automated tests on real devices, and send back reports that tell a developer how to reproduce each problem.
Get your testing scoped in writingWhat QA is
QA testing services put a piece of software through the situations its users will create and report what breaks. The work covers what to test and in what order, the tests themselves, and the reports that let a developer fix a problem without a meeting.
The test plan ranks each part of the product by what a failure would cost, so sign-in, shift swaps and payroll export get the deepest coverage and a settings screen gets a lighter pass.
Checks you will repeat on each release become automated suites in your repository. Anything that needs a human eye, such as a layout on a small phone, stays a manual test.
A bug report lists the steps, the expected result, the actual result, the device and a recording, so the person fixing it can see the problem on their own screen.
Two ways to test
The two differ in who knows the code and who is looking at it fresh.
| Build team tests its own workDevelopers or in-house testers | Separate QA teamA second set of eyes on the product | |
|---|---|---|
| Who knows the code | The people who wrote the code know where it is fragile and can test those parts directly | Testers start from the product’s behavior and the test plan, and read code when a report needs it |
| What gets found | Logic errors found while writing the code, through unit tests and code review | Problems a user would hit when following paths the authors did not think of |
| Whose code it can be | The team’s own code | Yours, another vendor’s or ours, since the testers have no stake in the code |
| Staffing | Permanent staff who learn the product in depth over the years | Brought in for a release, a project or a handover, then scaled down |
| Best when | A mature team has the time and the tooling to cover testing itself | A release is close, a vendor is handing over software, or no one on staff owns testing |
You may already have developers who write unit tests. Independent QA sits above that layer, with the browser, the phone and the user’s path through the product. If another vendor built your software, a written test report is also a record of what you accepted on the day you took it over.
Rules testing works inside
App store requirements, accessibility standards and the tools themselves. Sources under each.
Guideline 2.1(a) asks developers to test an app on-device for bugs and stability before submitting it, and to provide working demo account details if the app has a login. A release plan therefore includes a pass on physical iPhones and a demo account that works.
Google asks developers to keep the user-perceived crash rate under 1.09% and the ANR rate under 0.47% overall to maximize a title’s visibility. The limit for a single phone model is 8% on both, so one weak model can matter even when the overall rate looks healthy.
WCAG 2.2 adds success criteria to 2.1, and content that conforms to 2.2 also conforms to 2.1. Our accessibility pass tests against 2.2 level AA, which includes criterion 2.5.8, a minimum target size of 24 by 24 CSS pixels, with exceptions.
The Department’s web guidance says the ADA applies to services offered on the web. It names WCAG and the Section 508 Standards as existing technical standards that provide helpful guidance, and says the guidance itself has no legally binding effect. Our report records results against WCAG 2.2 and carries no legal opinion.
The WAVE engine scanned one million home pages, and low-contrast text appeared on 83.9% of them. WebAIM adds that no automated tool detects all conformance failures. For that reason a scan is one part of our accessibility pass, alongside keyboard and screen reader testing.
Playwright Test supports all three engines on Windows, Linux and macOS, locally or in CI, with mobile emulation for Chrome on Android and Mobile Safari. Playwright’s WebKit approximates Safari, and emulation approximates a phone, so the physical-device pass runs on real hardware in BrowserStack.
What we test
A scope picks the ones your release needs. The test plan says which, and in what order.
A written plan that lists the parts of the product, ranks each by what a failure would cost, names the method for each and sets the conditions for calling a release ready. You receive: the test plan and the case list.
A tester works through the product as a new customer would, then goes off the script with a question in mind, such as what happens when a phone loses signal halfway through a shift swap. You receive: the session notes and a bug report for each defect.
Browser and API tests written in Playwright or Cypress, and Appium tests for phones, committed to your repository and wired into your build service so a failing check shows on the pull request. You receive: the suites and the pipeline configuration.
Tests on physical phones and tablets in BrowserStack, chosen from your own analytics so the matrix reflects the devices your users hold. You receive: the device matrix, screen recordings of each failure and the OS versions covered.
Automated scans, keyboard-only walkthroughs and screen reader sessions with VoiceOver and TalkBack, each result tied to a WCAG 2.2 success criterion. You receive: the accessibility report, with the criterion, the screen and a suggested fix for each failure.
Scripts that simulate a stated number of users signing in, opening schedules and requesting swaps, run against a test environment, with response times and error rates recorded as the load rises. You receive: the load script and a results sheet with the point where it slowed.
How it runs
No stage goes live until you have approved it.
Test plan, release 4.2
| Area | Risk | Method | Cases |
|---|---|---|---|
| Sign-in and password reset | High | Automated browser tests | 14 |
| Shift swap and manager approval | High | Automated browser tests, exploratory | 22 |
| Payroll export | High | Automated API tests | 18 |
| Push notifications | Medium | Appium on real devices | 9 |
| Offline schedule view | Medium | Exploratory | 6 |
| Settings and profile | Low | Manual checklist | 7 |
Our stack
What the tests are written in, run on and read with. Pick one for the detail.
BrowserStack rents physical phones, tablets and browsers over the internet, so a test can run on hardware you do not own.
The device matrix is built from your own analytics, and failures are recorded on the real phone.
Questions
Keep exploring
Business software built around how you work, in your own repository.
Swift and SwiftUI iPhone apps, built in your own Apple Developer account.
Kotlin and Jetpack Compose apps, published from your own Google Play account.
Engineers on your product, working in your GitHub, Jira and Slack.
A remote team building in your repository, with the code in your name.
A first version at a fixed price, with a list of what to cut.
Portals, dashboards and SaaS-style tools that run in the browser.
Software for larger organizations, built around the systems already in place.
One TypeScript codebase for iPhone and Android, in your own store and Expo accounts.
Start here
Tell us what is being tested and who built it. We’ll send back a written scope.