Empirical Process Control
Empirical process control is the foundation of Agile Delivery. It is based on continuous improvement and visible progress. This means data from the past is used to plan for the future.
The resulting progress becomes visible, which helps in quickly reacting to new insights.
Value the last responsible moment
In an empirical approach options have value. Decisions remove potentially valuable options. Since every experiment can generate new insights that lets us develop a better understanding, we try to decide at the last responsible moment. Besides creating room for experiments the last responsible moment also enables us to react to shifting goals.
Know your business and customer focused goals
Before we can start working empirically we need to know what success and failure would look like. This means we have to know our goals. We recommend to start with a focus on the complete product from a business and customer perspective: What do our customers need and value? What quality criteria are important to them?
Validate assumptions through experiments
We formulate hypotheses about these goals and how to achieve them. Then we try to validate our assumptions with experiments. This creates empirical data that we measure and compare against our goals. We use these results for our next hypotheses and experiments.
Sector Rating Aspects
These are aspects that can be used to assess your performance and maturity in this sector. For a detailed explanation please have a look at our how to page.
|Process and progress data is visualized (e.g. burn-up / burn-down charts, Kanban Boards)||Late majority|
|Data from the past is the primary driver for forecasts of the future (e.g. velocity, throughput)||Late majority|
|Short iterations drive development (e.g. Sprints)||Late majority|
|Iterations have defined goals that are actively tracked||Early majority|
|Business representatives are directly involved in development||Early majority|
|Continuous improvement is an important driver (process, tools, communication are topics of retrospectives, not only business-related features are addressed)||Early majority|
|Categorize problems after their Cynefin domain to find out where iterative-incremental approaches have value (e.g. Agreement & Certainty Matrix Liberating Structure)||Early majority|
|Goals are used on different flight levels and interlinked (e.g. strategy/portfolio-focused, coordination/product-focused, operational/work-focused)||Early adopters|
|DevOps mindset is established (3 Ways of DevOps)||Early adopters|
|Fail fast, fail safe culture is established (small experiments, safe environments)||Innovators|
|The broad use of the last responsible moment creates learning windows (technical abstractions, re-prioritization of features, use of technical adaptors, …)||Innovators|
- Feedback & Transparency: Drive empirical process control through feedback – transparency is the precondition.
- Responsibility: Foster responsibility in teams and individuals to be able to identify and execute experiments and use the results sensibly.
- Verticality: Accelerate and simplify your experiments through decoupling and localization.
- (Guided) Architectural Emergence: Experiment in areas that are not optimally suited to do so through soft governance.