Услуга · Тестирование и QA

Test automation

Automation pays off where the same scenarios are checked over and over. With a yearly release it is unnecessary. With a fortnightly release the manual regression run becomes the main brake on the whole process.

The run
minutes instead of days
The build
checked automatically
Отбор
the key parts, not everything
Отчёт
what failed and at which step

What the work includes

We automate not everything but what repeats often and has settled down. Chasing full coverage costs more than the benefit it brings.

Discuss the scope

Selecting the scenarios

First of all what is critical to the business and gets tested most often.

Writing

The tests are made resilient to cosmetic interface changes, otherwise maintenance eats the entire benefit.

Test data

They are prepared and cleaned up so the runs do not affect one another.

Running on each build

The tests run automatically on every build rather than when somebody remembers them.

Reporting

A clear result showing the step at which it failed, with a screenshot.

Maintenance

Updating the tests as the product changes. That is continuous work rather than a one-off project.

How it goes

The first tests appear in two or three weeks; the noticeable effect comes once the main paths are covered.

01

Расчёт

We count how much time the manual regression run takes today. From that you can see whether automation will pay off.

02

The basis

The execution infrastructure, test data preparation, the reporting system.

03

Coverage

We write the tests in turn, starting with whatever breaking would be most visible.

04

Upkeep

We update them as things change and deal with flaky tests.

A flaky test is worse than no test. When a test fails again and again for no real reason the team stops reading the results, and genuine breakages get ignored along with the noise. Such tests are either fixed at once or deleted without regret.

Questions and answers

No. Automated tests run what has already been described quickly and tirelessly. Only a person can find an unknown problem in a new scenario. Automation removes the routine, not the people.

With infrequent releases, a rapidly changing interface and prototypes. If the product is redrawn every month the tests will have to be rewritten faster than they can pay off.

Sensible coverage rather than maximum. Usually the paths that bring in money and the places where defects have already occurred are enough. Chasing a coverage percentage produces tests for the sake of tests.

We will move the tests to automation

Tell us how often you release and how long the manual run takes. We will work out whether automation pays.