Andres Glusman is the Cofounder and CEO of DoWhatWorks. Prior to DoWhatWorks he led product and growth at Meetup where he was a pioneer in the lean startup movement. He has been running tests online since the late 1990s and is passionate about human behaviour. We spoke about the power of split testing, when you can do it, and when you can't.

Otherwise known as A/B tests, split testing involves systematically varying a user experience for different groups of users & then comparing the end result to see how those changes impacted behaviour for better (or worse).

These are terrible odds, but the good news is that terrible odds aren't hard to make a little better. It's important to get as much signal from as many sources as possible upfront so you have the best chance of success.

Split-testing is a great way to learn, but it's not the only way to learn. Make sure you use a mixture of experiments, surveys, qualitative inputs, feedback and any other data source you can get your hands on. It's all signal.

You need to be careful that you're going overboard. It's possible to test too much & get caught in eternal loops. It comes down to fear of mistakes but, unless you're putting medicine in someone, you can make a mistake.

You don't have to have a mass-market B2C app to get good tests going. There's scope to experiment in B2B but you need to do your homework, get as much data as you can upfront & make sure you put your chips on the right part of the table.

You can catch up with Andres on LinkedIn.