Pattern: Reduce Cost of Experimentation
When someone has an idea that requires validation, the costs of doing experiments around it needs to be as low as possible
A company is moving into cloud native with little understanding of tooling and technologies.Uncertainty is very high. New ideas and solutions come up all the time, andthe team has to run a variety of validation experiments to select the best ones.
In This Context
There are significant barriers to experimentation in the organization: permission is required, and the related planning, documentation, and coordination meetings take a lot of time. Then actually getting the results afterward typically requires a significant wait. As a result, engineers will often skip experimentation and move directly to execution.
- Very little chance that any experiment will lead to canceling a project because there is already high commitment to that path.
- Higher cost means an experiment is less likely to be done.
- If cost is low enough, people can experiment even with things that lead to failure.
- The higher the cost, the lower the willingness to abandon the experiment (sunk-cost effect).
- Traditional organizations require extensive documentation with any experimentation,which by definition makes costs high.
Therefore
Put in place a simple, straightforward, and seamless process for doing experiments.When experimentation is central to an organization’s process and progress,it needs to be an inexpensive and easily accessible action.Outline and publicize a process, or provide periodic training. for how to create appropriate hypotheses with measures and then design small experiments to test them. Aim to remove bureaucratic roadblocks such as managerial approvals and extensive documentation for every action, while providing technical infrastructure that is light, fully automated, and requires only a tiny budget.
- Create a framework around experimentation: the tools, facilitation techniques,project-management structure, and guidelines for allocation.
- Such facilitation makes it faster to obtain and understand the results of an experiment.
Consequently
More experiments take place. Instead of extensive research and guessing when a complex problem arises, a rapid process of hypothesis/results/analysis provides the solution.
- Teams can be more innovative when they know they can easily try out ideas.
- There is cost to experiments, even if it is low.
