This article explains how to design an experiment to help prove or reject your assumptions.

 

Before you can design your experiment, you need to have:

  • established the assumptions on which you have based your business model
  • categorised these assumptions against the four validation themes
  • assigned each assumption a level of priority

 

You now need to establish if these assumptions are valid. There is no one-size-fits-all experiment; your assumptions will be many and varied, and so experiments will also vary. In general, however, you should be aiming to gather as much information in the least time, and at the lowest cost possible.

 

We use experiment cards on Solverboard to help you collect all information relating to an experiment in one place, and link it to the assumption you are trying to prove. These cards encourage you to follow a simple four-step process:

  1. Define your method – What are you going to test and how? You should be testing something about which you are uncertain, so failure is to be expected.
  2. What outcome are you expecting? – We will ask you to pass or fail an experiment, so you need to have thought about what a pass looks like. What is your minimum expectation of proving the assumption?
  3. Capture learnings  – Whatever the outcome, every experiment provides another bit of knowledge that may be useful at some point in the future. Solverboard enables you to capture all the learnings, whether they are essential now or might be helpful at some point in the future.
  4. Declare the result – Objectively state the result: pass, fail or inconclusive.

 

Create an experiment

  1. Image showing adding an experimentIn the relevant assumption card click 'Add an Experiment'.

  2. Decide on an appropriate experiment depending on:
    1. what aspect of the assumption you are testing, for example, an experiment designed to test customer desirability may not work so well to test technical feasibility
    2. where you are in your testing cycle. Typically in the early stages of testing, you will want a fast indication of hypothesis confidence (and therefore cheap). As you proceed, you will need more certainty to your evidence which may take longer and cost more
    3. what constraints you are working with -  money, time and available skills will dictate what experiments you can run

  3. Give the experiment a title.

  4. Describe the experiment in terms of what aspect of the assumption you have designed it to test.

  5. Define the method of the experiment including any types of data you will measure.

  6. Describe what result you are expecting. What does the minimum acceptance of good look like?

  7. Assign the experiment owner and press save to publish your experiment.

 

Manage an experiment


Click the edit icon to make any changes and use the ‘Save’ button in the top bar to save.

 

  1. Image showing editing an experimentWhen you are ready to start the experiment in edit mode change the status to ‘in progress’ and save your changes.

  2. Conduct the experiment and gather as much evidence and data as possible. You can add files using the attachments section.

  3. Record the outcome achieved in ‘Actual result’.

  4. Depending on your outcome, update the status field to ‘valid’, ‘invalid’ or ‘inconclusive’.

  5. Save your changes.
     
  6. Capture learnings. Click ‘Add a lesson’ to record all learnings to support the experiment’s conclusion and to transpose into your initial business case.

Stuck? Want to talk to us?

Raise a support request or send us an email