A weekly execution review that measures closure instead of busyness

Define the commitments before calculating the score

Begin Monday by naming the commitments you intend to complete that week. Record their owners, completion conditions, and decision dates. Preserve that cohort for Friday's review even if new requests arrive. You can still accept new work, but mark it as added after planning. Otherwise an expanding backlog makes the closure denominator move underneath the team.

Critical Path' current Analytics period filter is not a strict weekly cohort. Source inspection on September 21, 2026 shows that open work remains included even when older, while completed or cancelled records are scoped using completion, due, or first-history timestamps. The owner filter narrows tasks, but the outcome-health portion of the operating score uses active workspace outcomes. Do not describe the composite as a precise personal weekly productivity score.

For a rigorous review, keep a short private cohort note alongside the dashboard. Write the period, owner set, number of planned commitments, additions, cancellations, and missing records. The note is small enough to maintain without creating another reporting project, yet it prevents most arguments about what a percentage really means.

Use a definition ledger rather than trusting labels

The table distinguishes current implementation from the decision you should make. These definitions are based on our application source, not a certification of measurement quality. A metric can be useful while still requiring context. In particular, absence of data must not be interpreted as proof that a process is healthy.

Work through a fictional ten-commitment week

A fictional operations team begins with ten commitments. By Friday six are completed, one is cancelled after a deliberate scope decision, two are waiting for external input, and one remains in progress. Three additional requests arrived midweek; one was completed and two remain open. The original-cohort completion share is six of ten, or 60 percent. The final thirteen-item set has seven completions, roughly 54 percent. Both numbers are arithmetically correct and answer different questions.

Cancellation deserves its own explanation. It may be excellent prioritization rather than failure. If you also report closure excluding cancelled work, label it explicitly: six of nine retained original commitments, about 67 percent. Do not silently switch to this more flattering denominator and compare it with last week's original-cohort measure.

Suppose four of the six original completions had explicit deadlines and three met them. The dated-cohort on-time result is three of four, or 75 percent. The two undated completions should not inflate that particular measure. Keep the underlying counts because a single changed record moves a small percentage dramatically. This example is invented to illustrate the method, not a customer result.

Interpret time measures without inventing hours saved

The Average completion card measures elapsed hours from a saved creation event to completion for actions completed in the selected window. Later notes do not change it. Missing or inverted timestamps produce an unknown result, not a negative duration. It is not a deadline offset or measured active work duration. Provider acceptance separately counts accepted and failed send attempts in its selected window; retries count again, while scheduled and skipped reminders do not. Acceptance does not confirm delivery to the recipient.

Similarly, the current chart captioned Completion by quadrant uses active, non-completed, non-cancelled task counts. Read it as the distribution of open work, not evidence of which quadrant produced the most completions. That distinction matters when a growing strategic backlog looks like progress merely because its segment grew.

For protected time, inspect an actual provider-synced block when external reservation matters. The setup checklist requires a connected calendar and an upcoming or in-progress block with provider identity and sync evidence; a general planned-minutes label is a different signal. Estimates may also come from defaults when explicit effort is missing. Record which commitments were genuinely estimated before comparing planned load with available time.

Run a twenty-minute review that ends in decisions

Treat twenty minutes as a suggested meeting budget, not a measured best practice. If the review expands, reduce the number of metrics before compressing the discussion of important exceptions. A small team normally needs to know what closed, what changed, what is blocked, and which commitment now needs a different decision.

Review evidence at the action level before changing statuses. A completed calendar session is not necessarily a completed deliverable. A waiting action should name the dependency and next review point. A blocked action should identify what would unblock it. Without those definitions, the team can improve percentages simply by relabeling work.

1. Confirm the review period and fixed cohort. Separate new arrivals and cancelled work.
2. Read each completion condition and identify the result produced, not merely activity performed.
3. Inspect late and unfinished commitments. Name the constraint: scope, estimate, dependency, ownership, or unavailable time.
4. Check calendar reservations and a small sample of integration evidence where a missed block affected execution.
5. Choose at most three changes for next week, assign an owner, and record when each change will be reviewed.
6. End by reducing or renegotiating commitments that exceed realistic capacity. Do not raise the budget just to make the chart greener.

Use the operating score as a prompt, not a verdict

Critical Path currently averages five components equally: completion, reminder send-state share, delegation completion, inverse deferral rate, and outcome health. Empty component sets have different defaults: for example, no active outcomes gives full outcome-health credit, while no reminder or delegation records gives zero for those components. The resulting number is a heuristic, not a validated measure of organizational performance.

A team that legitimately delegates nothing should not invent delegated tasks to improve the score. A team that needs fewer reminders should not create more just to change the ratio. Review the components relevant to the work, note non-applicable ones, and prioritize observable failures. The point is to improve decisions, not optimize the display.

For outcomes, verify the real success condition outside the task count. Closing all preparation actions does not establish that a customer renewed or a hiring decision was correct. Keep operational closure and business impact related but separate. Where evidence is incomplete, write unknown rather than replacing uncertainty with a confident percentage.

Keep a private, reproducible review record

Use the weekly operating-review worksheet to capture the cohort, evidence, and decisions. A useful record has enough detail for next week's owner to understand the change, but not a dump of every confidential conversation. Do not publish raw task titles or employee rankings to make the review look objective.

Critical Path' Export snapshot downloads JSON containing the current workspace state and, when present, pending recovery information. It is not an anonymized analytics report or a guaranteed one-click re-import. Store it privately and inspect it before sharing. If a save conflict is visible, preserve the recovery copy before loading the server version; do not treat the downloaded file as proof that unsaved changes reached another device.

The best evidence of an improving review is a specific repeated failure becoming less frequent under a stable definition. Record the counts, the intervention, and alternative explanations. A busier week with a lower completion percentage may still contain better decisions than a quiet week of easy tasks.

Common questions