Stop A/B testing before you have traffic

[Invented] A founder asked us to A/B test two homepage headlines. The site got about forty visitors a week.

[Invented] Do the math. Forty a week is roughly a hundred and sixty in a month. Split that in two and you've got eighty people per version and maybe three clicks. That isn't a test. It's a coin toss with a spreadsheet.

[Invented] Testing is a tool for when you have traffic and a hunch you can't settle by thinking. Before that, it's a way to feel productive while avoiding a decision.

[Invented] So we picked the headline we believed in, shipped it, and spent the time talking to the people who did show up. Five conversations taught us more than a month of split testing would have.

[Invented] Early on, don't test. Decide, publish, listen. Save the experiments for when the numbers can actually answer you.