Learning Versus Throughput Experiments

I broadly divide experiments into two categories:

Learning experiments and throughput experiments.

The difference between these experiment types is the goal.

Even though the build-measure-learn loop ends with learning,

The goal of a lean startup isn’t just learning,

But turning learning into measurable business results – aka traction.

The goal of every experiment should be testing

a bigger strategy for increasing traction

by running throughput experiments

unless we run out of potentially good strategies to test.

Then we fallback to running learning experiments

as a way to generate new strategies

which we then test with throughput experiments.

Examples of throughput experiments:

• Solution interviews

• Teaser pages

• A new feature launch

Examples of learning experiments:

• Problem interviews

• Usability tests

• Surveys

Learning experiments are for hypotheses generation while

throughput experiments are for hypotheses validation.