Explore vs Exploit

Well, this tweet got some traction:

The goal of testing a lot of growth ideas is not to find a lot of things that work.

The goal is to find your points of leverage as quickly as possible and focus your limited resources.

— Brian Balfour (@bbalfour) August 26, 2021

Let me expand. There are two modes of experimentation:

1. Explore - Exploring is about testing wide to find your points of leverage in your growth model so that you can focus your limited resources and exploit.

2. Exploit - Once you find our points of leverage, you move into exploit mode by testing deep and narrow.

Knowing which mode you are in is key. There are some common mistakes I see:

1. Moving Into Exploit When You Are Better Staying In Explore

   When you are in explore mode, you are testing wide but you aren't necessarily looking for a lot of things that work. You are trying to learn about a small set of areas that huge ceiling room for you. A point of leverage. This requires discipline because you might find some thing that works, but move on from it. With your limited resources those resources could be better spent staying in explore mode to find a point of leverage vs moving into exploit.

2. Moving Into Explore When You Are Better Staying In Exploit

   The flip side also happens. Teams stop exploiting a certain area in their growth model, and move into explore. This happens commonly for two reasons:

   • Teams underestimate the depth in a certain area. Dan Hockenmaier (Reforge Partner and Head of Strategy at Faire) and Lenny Rachitsky have a good framework around this they call Validate, Commit, Scale. They say most teams have trouble to committing because it involves:

     "Dedicating a significant amount of cross-functional resources to the effort, including product, design, marketing, and engineering

     Influencing the core product roadmap and customer experience to optimize for the lane being pursued. "

   • Teams get bad advice (often from investors) that they should "diversify." Diversification is the enemy of leverage.

Related Links: Good Experiment Bad Experiment, Experiment Management System