Martin Staal Boesgaard

Good test data is essential to all stages of testing. If the test data does not represent the reality of data in production, the test itself does not represent how the tested system will behave in production. We should extend our view on who is consuming test data. Developers can also benefit from representative test data, and the same goes for users/customers/vendors/partners involved in the development process. Furthermore, we should keep in mind that different stages of testing also have different requirements to test data. Some may even require hybrid approaches. There are many methods of generating test data. In most applications, multiple methods must be deployed in order to achieve the quality and properties needed. The different methods are discussed with pros and cons. If test data is derived from production data, privacy, GDPR, and security must be taken into consideration. Using production data for testing is not legal. Cost-efficiency and how to integrate and automate test data generation will be discussed in order to make test data generation become a natural step which does not consume too many resources.