The reference analysis then produces a number of hypotheses that you want to test with a test. Ideally, you use A/B testing for this via tools such as Visual Website Optimizer , Unbouncy or Optimize used at WelkeKoffer . A rule of thumb of 1,000 visitors Then you start testing. If a test is significant, you analyze the data and see if the stated hypothesis can be confirmed. for this obviously varies. thumb of 1,000 visitors. However, it strongly depends on how much difference there is between a test. The more the test variants differ, the sooner a test is significant. Signage can therefore occur sooner than 1,000 visitors in some situations. Finally, you make the changes and continue with the next test so that it becomes a continuous process.
This way you keep working on improving the performance of your affiliate website. In summary, your method could look like this: Analyzing the data (references) Make the phone number list same hypotheses Testing hypotheses with tests Analyze data Implement changes 'On to the next one' Then on to a number of tests that we have done that are worth mentioning and from which there is some inspiration. Of course, not all tests were positive. In the case of a test that is negative, valuable information can often be obtained. For example, the tests with the subject a certain position in the top 3 suitcases on the right side of the homepage, often turned out negative.
Don't be alarmed by a negative test A change on the website will only be implemented if the new test variant is positive. If this is not the case, of course you leave the situation as it was. This is also another nice functionality of the tool that is used. This is because no 'drastic' interventions are required from the ICT department during testing. Test variants can simply be created in the tool itself. By the way, negative stands for loss in commission, CSR and eCPC. With the results of a negative test you often have input for new tests. So don't be put off if a test is negative, but learn from this.