Clean Faillure

As a nice play on the name of this site, here is an article about a clean startup :-) The article talks about Homejoy and why it failed. Ultimately, they never really validated their #solution.


And if you like more, here is a discussion on Hacker News:


Validation Pyramid

A sequence of 5 steps for validating business ideas/models. It is named pyramid, because each step builds upon the steps before. 

The steps are:
  1. Customer Segment
  2. Relevant Problem
  3. Accepted Solution (Problem-Solution-Fit)
  4. Viable Channel
  5. Monetization (Product-Market-Fit)

Whenever a step is validated, all steps with a smaller number are validated as well. E. g.: if your customers can imagine to use your prototype in reality and pay money for it, they must have a relevant problem.


Lean Startup

A process for systematically reducing risks in innovation projects.
The core idea of lean startup is to minimize risks by actively collecting data for risky parts of your plan. Lean startup is a process for any risky projects, not just startups.



Hypotheses can be validated. This is done by designing and running an experiment, then interpreting the results. If the data does not falsify the hypothesis, the hypothesis is validated. The main effect is, the probability of an assumption goes up (which you hope) or down (which you fear, but should know the sooner the better).



An experiment is a way to collect data, which could falsify a hypothesis (H). If the amount of data is so large that it could falsify the hypothesis, but it doesn’t, then the hypothesis is considered true. Or: How much data do you need to convince yourself, a team mate, or a VC of H or its contrary?
Most experiments measure user behavior, for which customer conversations is the main technique. Others kinds of experiments are data research and technology tests.


  • Go to the central cafeteria of the university and ask 20 students how much they spent on their bike-lock.


Hypothesis (H)

Each assumption can be formulated as falsifiable hypotheses = there exists an experiment, which proves or disproves the hypothesis. Formulating good hypotheses is an art.

  • At least 20% of students paid more than 100 € for their bike lock.