Engineering delivery metrics that explain blocked work.
Define delivery-time boundaries and show unfinished work beside completed-work statistics. Review blocker causes and next actions with the client and engineering team so metrics improve shared decisions rather than become individual productivity scores.
Inside this article 6 sections +
Engineering delivery metrics should reveal where work waits, what remains blocked, and which decisions would improve flow. Define lead-time and cycle-time boundaries, show active work as well as completed work, and review the reasons behind long delays.
For a dedicated engineering team, this visibility helps the client and delivery team share responsibility. A blocked requirement, missing environment, or delayed review is a system condition to resolve, not a number to use for ranking individual developers.
Define when work starts and finishes
Choose a work item meaningful to the delivery agreement, such as a customer capability or a reviewable implementation slice. Record when it is accepted into the delivery commitment, when active work begins, and when it meets the agreed completion criteria.
Call commitment-to-completion lead time and active-start-to-completion cycle time only after naming those boundaries. Other frameworks use the same terms differently. DORA's change lead time, for example, measures code commit to successful production deployment; it should not be presented as the same interval as a product request waiting for clarification.
Use consistent clocks. Calendar hours, working hours, and working days produce different results. Select the view that supports the decision and label it. Do not subtract weekends from one item and leave them in another simply because the resulting chart looks smoother.
| Work item | Committed → started → done | Lead / cycle | Blocked within cycle |
|---|---|---|---|
| A · export filter | Hour 0 → 24 → 48 | 48 / 24 hours | 4 hours: example clarification |
| B · account selector | Hour 8 → 12 → 36 | 28 / 24 hours | 12 hours: review dependency |
| C · provider connector | Hour 0 → 24 → still open at 60 | Age 60 / active age 36 hours | 20 hours: sandbox access |
Show active age beside completed-work time
Completed-item charts describe work that finished. They can look healthy while the hardest items remain open indefinitely. Add the age of active work and the age of its current blocked state, with a named next action.
Separate total age from blocked duration. An item may have spent time in several states, and a blocker can begin after productive work has already occurred. Keep the state transitions or a lightweight event log so the team can explain the interval rather than relying on a manually edited total.
For the synthetic ledger, the two completed cycle times average 24 hours. That statistic says nothing about the connector that has been active for 36 hours and remains blocked. Showing both views changes the review from celebrating a neat average to deciding how to obtain sandbox access.
Classify blockers by the decision they need
Use a small set of useful causes: missing product decision, external access, cross-team dependency, review capacity, or technical investigation. Include the owner of the next action and the date it was requested. A label reading “blocked” without an action merely documents delay.
Distinguish uncertainty from inactivity. A bounded technical investigation may be legitimate active work if its hypothesis and expected evidence are clear. If the investigation has no decision boundary, it can become a hidden queue of unresolved design questions.
Review recurring causes across items. Several delays waiting for the same environment suggest an access or provisioning improvement. Several delays waiting for product clarification may suggest that work enters commitment before acceptance criteria are ready.
Use work in progress to protect focus
Count started but unfinished items at a consistent level. Splitting one feature into many tiny tasks changes the count without necessarily changing the team's real workload. Keep the work-item policy stable enough to make trends interpretable.
When many items are blocked, starting more work can hide the problem and increase context switching. Review whether the team should resolve a dependency, reduce scope, or finish a nearly completed item before beginning another. A work-in-progress limit is useful only when the team agrees what happens when it is reached.
Do not infer a universal productivity target from these counts. Different items carry different uncertainty, review needs, and customer impact. Use the metric to identify where a conversation is needed, then inspect the work.
Connect flow to scope and cost visibility
Show material scope changes separately from time spent waiting. If an item gains a new provider integration halfway through implementation, its longer duration should not be interpreted as unexplained slowdown. Preserve the original commitment and the decision that changed it.
Connect the delivery review to a forecast with explicit assumptions. A small completed-work sample is not enough to promise a fixed completion date for a different workload. Explain what dependencies remain and how resolving them would change the forecast.
The scope and cost visibility guide helps keep those decisions understandable. The asynchronous status guide provides a communication format for the evidence behind the numbers.
Run a review that produces actions
Review the oldest active work, the oldest blockers, and a few completed items with unusual waiting patterns. Decide who will resolve each dependency and when the team will revisit it. Keep the meeting focused on changes that improve customer delivery rather than explaining every small fluctuation.
Use the same ledger across the client and engineering team so neither side must reconstruct context from private task boards. Dedicated teams work best when responsibilities, blocked decisions, and delivery evidence are visible together. These metrics should make that partnership easier to operate and the next useful action easier to see.