Complexity and Estimation

When refining work, teams should focus on reducing complexity, minimizing dependencies, and estimating based on complexity and effort, not time.

Small things can be estimated more accurately than big things because the margin of error is lower and dependencies are clear. Eliminating or reducing hard dependencies is critical because the probability that something will be delivered late doubles for every hard dependency. Those could include database changes, coordination with other teams, or changes that are tightly coupled with another component.

Recommended Practices

Decompose stories using Behavior Driven Development. This not only helps with feature discovery and with uncovering dependencies, but also aids with story slicing since each acceptance test is naturally a thin, vertical slice.

Prior to refining, use relative sizing to project order of magnitude estimates for delivery. However, these should not be used for commitments. Committing to unrefined deliveries increases team allocation with re-work and “Date Driven Development”, reduces quality, and lowers end user satisfaction.

To avoid hard dependencies, first slice stories as small as possible to minimize the number of possible dependencies. After that, attempt to make any hard dependencies “soft” with feature flags, API versioning, or other coding solutions.

Tips

• Use Cynefin to aid in estimating complexity.
• If the team does not agree with the estimate, refine further. Avoid “averaging” the team’s estimate.
• Track estimates against actual to see how consistent the team is.