Lessons learned from Paul Willard's talk on Growth Hacking

A few weeks ago, the free book, The Definitive Guide to Growth Hacking, traveled around Internet like a scalded troll (Swedish expression for viral). At the end of the book, the author of the book, Neil Patel, argued that the company Atlassian is a role model on how to grow a company. The book said:
Atlassian is one of the fastest growing software companies in the world and they also have one of the most sophisticated growth hacking teams that I've ever seen. They approach their experiments with the rigor of a scientist, and they use machine learning models to refine their process over time. They have one group focused specifically on new customer acquisitions, and another group focused exclusively on the funnel once they are inside the product. It's a well oiled machine. 
But here's the rub; they are an enterprise B2B software company, selling products that the general public has never heard of. They have boring products that enable communication and collaboration within software environments. They are not spreading because they are sexy. They're growth hacking their way to the top. If you ignore Atlassian, and the companies that are following their lead, then good luck competing in the future.
So I thought it was a wise idea to research exactly what Atlassian are doing to be such a role model. It turned out that the Chief Marketing Officer at Atlassian is Paul Willard (Twitter, LinkedIn, Blog) and he has given a lecture on the subject.

Lessons learned
  • He learned the basics of growth hacking from NextCard in the late 1990's
  • He holds an M.Sc in Manufacturing Systems Engineering from Stanford, where he learned Japanese manufacturing concepts (Lean), and he can use those concepts when growth hacking
  • The saying "If you build the best product, you will win" is not true - the product will not travel around the Internet like a scalded troll. You also have to learn how to fail fast (2 month per A/B test is too slow)
  • Temporarily change what you want to test front-end since it's easier. If it's working, you can change the back-end code
  • For each person in the group working with growth, try 2 new ways each week to bring in new customers, while at the same time you continue with the old ways of bringing in new customers
  • A couple of tests each week per person is enough
  • Business changes, so what didn't work a year ago might work today
  • Read everything on behavioral design by Robert Cialdini, Dan Ariely, and Nir Eyal (Blog)
  • Speed of the website, call to action buttons, and grouping/contrast of the website are important
  • Don't A/B test your core repeat usage product (the customer will become annoyed) - it's better to use beta in that case. Only A/B test newsletters, tutorials, first-time user signup, etc
  • Keep things simple
  • An A/B test can also teach you about your market/segment