2 minute read

Tal Raviv’s article, That’s Not a Hypothesis describes what hypothesis is in the scientific method: something we believe in, a part of the model what we believe and something that can be falsified. This is connected to problem statement, the hypothesis is our belief why the problem is happening. If the hypothesis is true, we can make a prediction what will happen.

This structure is very similar to the Opportunity solution trees.

Setting problems, beliefs, bets

First, we start of a business problem. This is something we observe in our systems, or in the world that we think is worth solving. “We observe…”.

  • Business problems should be understood. If it’s data, there needs to be some exploration what it means, user research to figure out how does it affect our user’s behavior, market research to understand what other companies and the broader industry is doing to address the problem.
  • This is the first half of the first diamond.
  • End result should be a list of user problems we identified, that connects to the business problem.

Second, we write a list of hypothesis, beliefs based on what we know at this point. Our knowledge is still not certain, remember we want to solve the problem and still need to figure out the right way to solve it and also solve it the right way. “We believe that (assumption) … because (evidence) …”.

  • Beliefs should be about user problems and a list of supporting evidence (why we think these user problems are true). This is the most important part, as the solutions should be ultimately about solving user problems that drive the right business outcomes.
  • Based on this list we can prioritize which of the user problems we found can we also solve, and seems we can come up with effective ways of solving.
  • Based on this list we can clarify the design intent, what is the problem and constraints of the problem we will work with.
  • This concludes the first diamond.

Third, based on the beliefs we can create a list of bets for each beliefs, possible ways that reflects on our beliefs and ultimately solves the initial problems. “If we (do this) … We’ll observe (result) …”.

  • Each bet is a possible solution to the users’ problem that we also think would drive the initial business problem.
  • Since we don’t know yet which bet has the best potential (in the most rough sense for example impact / effort), we need to learn about each of the bet.
  • Sometimes a list of small impact / small effort bets are better than a large one, but usually it should be clear which list bet(s) are the most lucrative ones.
  • To learn about a bet, further discovery is needed - this mostly means explorations, experiments, anything that lets us learn about the bet and the belief, and helps us prove, we made the right the prediction with the bet.