If you want to improve your conversions you have to run A/B tests. But how long should you run them for?
The first thing you have to look at is statistically significant. If you are using tools like Crazy Egg, it will tell you which variation of your tests is performing better and by how much.
Once you have a 99% statistical significance, you can then make the winning variation the default.
The second thing you have to do is run your tests for at least a week. Traffic that you get after working hours or on weekends isn’t the same as peak hour traffic. For that reason, you need your tests to be live for at least a week.
Lastly, you need at minimum of 100 conversions per variation before you decide which one you are going to declare the winner.
Now, when following these tips you can’t just follow one, you have to follow them all of them if you really want your conversions to go up.