When incremental experiments plateau, step back and consider a complete overhaul to reach a higher local maximum
Continuous small experiments can lead to a local maximum where further optimization yields diminishing returns. Recognizing when you've topped out on a small hill and need to descend before climbing a bigger one requires stepping back from the day-to-day experimentation mindset.
When to use
When growth metrics have plateaued despite many incremental experiments and optimizations over an extended period
Don't do this
Continuing to run small experiments indefinitely, never questioning whether the fundamental approach needs rethinking
2 Founders Who Did This
After hundreds of incremental experiments on onboarding couldn't move growth metrics, Weiss led Project Day One - a complete overhaul that scrapped everything from website landing through first conversations
After 4 years as a passive directory at ~$2K/mo, completely overhauled Uneed into a launch platform with queue system, rankings, and badges