Which test cases to automate first
Not every test case should become a test. We rank them by two things: how much it costs you if the case fails in production, and how often someone runs it by hand. The cases that are high on both go first.
Some work stays manual:
- one-off checks that will not be repeated,
- checks that need human judgement, such as whether a page looks right,
- exploratory testing, where a person looks for what nobody wrote down.
That work is covered by manual and exploratory testing.
You may be in one of three situations. You repeat the same manual checks before every release. You have considered automation but never started. Or you automated the easy cases, and the complex ones are stuck. The audit starts from whichever it is.
Test case automation: from test case to automated test
Test case automation follows the same steps for every case:
- Read the case or the story and agree the expected result with someone who knows the business rule.
- An agent drafts the test from the case, in your framework.
- An engineer completes it and checks that the assertion is the business rule, not just that the page loaded.
- Pull request review in your repository. Nothing is merged without it.
- The test runs in your CI. How often it runs is agreed with you: heavier tests on a build or on a schedule, lighter ones more often.
For tests that run as part of your pipeline, see CI/CD test automation.
Your framework, or one chosen with you
In test automation development the framework comes first, because tests follow their framework and cannot be moved freely to another one. If you have one, we write in it. If you have none, the audit compares options for your stack, for example Playwright and CodeceptJS, and the report records the choice with the reason. If you need a framework built or reworked first, see custom test framework development.
Zapple.Tech was founded by the founders of CodeceptJS and Codeception, open-source testing frameworks, and of Testomat.io and Testomat.ai. So the conventions your new tests follow come from a company whose founders build test frameworks and test management. If one of those tools is an option for your stack, the audit says so and judges it by the same criteria as the others.
What makes an automated test worth keeping
A test earns its place when the next engineer can read it, trust it and change it. The tests we write follow a few rules:
- One behaviour per test. When it fails, you know what broke.
- Test data is not hard-coded. The test creates or loads what it needs, so it does not depend on what a previous run left behind.
- Names say what is checked. A failing test name should tell a developer the rule that failed.
- Locators are stable. Tests use identifiers that survive a change of layout, agreed with the developers where the product needs them.
- A test can fail. A test that cannot fail is not done.
The agent writes the draft, the engineer writes the assertion
An agent is quick at the routine part: a first draft of a test from a case, filling in the steps. It is poor at knowing which business rule the case stands for. A test that passes while checking the wrong thing is worse than no test, because people trust it.
So an agent drafts, an engineer decides what the test must prove, and each test is reviewed in a pull request before it is merged. The agent-driven way of writing browser journeys is described under end-to-end testing with AI agents.
What the work covers
Automated test scripts are written for the platform you ship on: web application testing, API test automation and the rest of the platform pages under test automation and QA testing services. Regression suites are a test type with their own page: regression testing.
When test script development is the wrong choice
- You want the whole automation project handed over, from strategy to a running suite. See test automation outsourcing.
- Your existing tests keep breaking. Repair is a different job from writing new tests. See test automation maintenance.
- You have no framework and no idea which one. Start with the automation audit.
- You want engineers inside your team, not tests delivered. See hire test automation engineers.
