The design of the web elements’ selectors is one of the most important design features of the front-end test automation. Although it might sound good, the maximum possible stability of selectors is not desired. The problem of a too stable selector is that it describes an element even if its substantial characteristics have changed. What are "substantial characteristics"? The only reasonable answer, covered in this talk, is that these are the ones determined by the scenario for the test automation—or by a common sense when the former is missing. A short example: users/testers are used to seeing a "Continue" button in a certain place in an application. Then a sudden change in the application renames the "Continue" button to "Next". Are the users/testers going to be confused? Yes, they are. And should the test handling the "Continue" button, which simulates users’ behaviour, fail? It should—but it won’t if the button is referred as id=cont_btn instead of its title.