Test early, test often

Testing is basically for free. If you are spending money on acquisition anyway, you might as well craft  multiple versions, and discover which ones work best.Test the big things first. Test for changes that—if successful—can really move the needle.

Test giving the paid product away, for free. Test completely different landing pages. Test doubling the price, or cutting it in half. Always be testing, and test the big things first.


  • Some channels facilitate rapid testing and iteration better than others. For instance, it is much easier to test 10 different variations of an online ad, then it is to test various content marketing approaches. For that reason, many people benefit from running paid ads in the early days to facilitate quick testing (law #4).
  • Traffic is always higher at the top of the funnel than at the bottom. To get conclusive tests, you need a minimum amount of transactions to draw conclusions. It’s much faster to get conclusive results at the top of the funnel (e.g. “more people clicked on advert A versus advert B”, than down at the bottom of the funnel. Start testing at the top, and then work your way down.
  • Don’t optimise small changes (like a landing page button copy), before you have optimised the big pieces (e.g. ask people to buy directly, versus capturing leads first). Validate the big structural pieces first, and test details later.
  • Don’t mix your variables. Learn how to run clean tests, with one variable changing at a time. If your new variant beats the baseline, make sure that you actually know what caused this improvement!
  • Once you find something that works, we recommend startups to run approx 70% of their traffic to their “baseline funnel” of their current best performing funnel. The other 30% of traffic goes towards experiments that are aimed to beat the baseline.
  • Only run experiments that you truly believe can beat the baseline. Even if you do so, only a small percentage of tests (typically 10-20%) will meaningfully improve your baseline. But this is a winning strategy: you can drop the losers, but keep the winners indefinitely, and continue building on top of them!

In Practice

Sean Ellis, the marketer that originally coined the term “growth hacking”, spoke extensively about how he used a process of high-tempo testing to drive growth at various companies, including his own online platform, It’s all about process — a steady pipeline of experiments being rolled out across the entire funnel is the most dependable way to unlock faster growth. It also conditions your organisation towards being data-driven, adaptive, and constantly looking for improvements.

Pair with

Optimise for tempo