Spot product or channel fatigue, acknowledge new competition, make needed updates in a timely fashion.
A good example to illustrate what doubling down on successful layers means is GrowthHackers. The launch of their email newsletter, which highlights the top posts of every week, was an important early success in driving readership. Yet, they didn’t declare victory and hang up their hats, they experimented with moving the signup form for the newsletter from the bottom of the landing page to the top, which drove a 700% increase in sign-ups. And they didn’t stop there.
Invest time and money into reviewing each of the major tasks and pathways your users take to reach your aha moment, to identify gaps in data and fortify it.
For example, Facebook’s growth team were coming up with a dwindling number of ideas for driving growth. In response, they ceased all experimentation for a full month and invested in improving their analytics tracking. This allowed them to perform more refined and powerful data analyses.
If the growth team has been working without a dedicated data analyst or data scientist, you should consider hiring or moving one over to work with the team full-time.
Experiment with adding new channels, if your team has reached an impasse where additional ideas are hard to come by within existing channels.
For example, for a company that has been reliant solely on paid acquisition, the growth team could experiment with developing organic channels such as search engine optimization, content marketing, or social media marketing to complement the paid efforts.
For example, Ankur Patel, principal growth hacking lead at Microsoft, regularly brings together product managers, engineers, and designers from different teams at Microsoft to share new thinking and insights to spur bursts of new ideas for his team to test.
Break out of the bounds of currently successful ways of operating by testing significant redesigns of features, company’s marketing, or products that have been successful, to see if they might not be substantially re-envisioned.
Start relatively small by challenging features or screens that seem to have been optimized, to see whether they might be even more effective if overhauled. Think of this as the principle of pushing past the plateau of a local maximum. A local maximum is defined as the highest value in a current set of points, but not the highest point overall.
For example, running experiments on the same pricing page over the course of a year can lead you to a local maximum for the performance of that pricing page. Regularly schedule experiments that go beyond optimizing to innovating bigger changes.