This lesson shows you how to set up your experiments so you don’t need to monitor their stats.
How is this possible?
There is a number of clicks (you’d calculate this threshold during one of the previous lessons) when your numbers become statistically significant.
- If you peek at stats before this threshold is reached, you get statistical noise.
- If you miss this point, you may lose money on an underperforming ad. By the way, your stats will NOT get any higher significance if you run them for a longer range!
In this lesson you’ll learn how to set up experiments so that a system makes the following actions when the threshold is met:
- Stops underperforming ads in time automatically (yes, you can have peace of mind, even when you launch experiments on Friday night 🙂)
- Notifies you when the test ad performs better than the control ad. (You can even set it up so the system will pause the control ad, but it’s not recommended. You’ll learn why that is in the following lesson.)