This message was deleted.

Hi. There are a few common strategies to testing an experiment before running live traffic at it. 1. Clone the experiment for use in a dev/staging environment; change the tracking key. This is probably the cleanest option. 2. Use a separate experiment phase (phase 1 = QA, phase 2 = live experiment) 3. Wipe the QA data in your datasource before starting and/or use a separate datasource for QA