Cycle time dashboard: delays and bottlenecks

A cycle time dashboard compares the duration of completed work with the age of work still open, using a stated start, finish, and clock. It should show where time accumulates and which cases need attention. A falling completed average is incomplete evidence if the oldest cases remain unfinished.

Define the clock before choosing the display

Choose a business event for each boundary: request accepted, processing started, and work completed, for example. State whether elapsed time includes nights and weekends, which time zone applies, and whether a reopened case keeps its original start. A target is meaningful only when measured on the same basis.

This example uses calendar days, with every timestamp recorded at 9:00 a.m. Eastern Time. Intake runs from acceptance to handoff; processing runs from handoff to completion. The two stages cover the full interval without gaps or overlap.

For a specific commercial process, use the definition in booking-to-invoice cycle per customer. Delivery reliability has its own boundaries in how to calculate OTIF; do not borrow its denominator for a general duration review.

Arrange the dashboard so each panel answers a distinct question:

Keep the cohort and snapshot visible. A completion-period chart and an open-work snapshot are different populations and should be labeled accordingly.

Work the completed cases from timestamp rows

A fictional US commercial-services team completed these cases from October 1 through October 14, 2026. The dates below abbreviate October dates; all are in the same month and year and at the stated time.

For timestamp columns B through D, the spreadsheet calculations for row 2 are:

Intake days:     =C2-B2
Processing days: =D2-C2
Total days:      =D2-B2

Store actual date-time values, rather than text labels such as the abbreviated display above. Format the results as numbers. Dividing elapsed hours by the hours in a day would be an alternative convention, but do not mix units within the same review.

The dashboard's completed measures follow directly:

Displayed means and percentages are rounded to one decimal. The NIST explanation of measures of location distinguishes the mean from the median and explains why skew can pull the mean toward a long tail. Case F is the long completed case here, so the difference between 5.7 and 4.5 days is useful evidence to investigate.

Keep the open queue beside the completed view

At the October 15, 2026, 9:00 a.m. snapshot, these cases remain open:

Open count is three, and average total age is 30 / 3 = 10.0 days. Case H is the only current-stage age greater than five days. Case I is the oldest overall, even though its current-stage age is shorter than H's.

The review therefore needs both fields. The processing owner investigates H's current delay; I's owner examines its earlier history and remaining work. Do not replace either age with zero because the completion timestamp is blank.

Change one delay and test the result

Suppose case F's processing time changes from ten days to four, with every other input unchanged. Its total duration becomes 3 + 4 = 7 days. This variation illustrates a shorter stage; it does not establish what operational change would produce it.

The mean falls because 28 / 6 is lower than 34 / 6. The median stays at 4.5 days, and target attainment stays unchanged because F still exceeds five days. The open-work snapshot also remains unchanged. This is why a dashboard needs more than one summary figure.

Checks and pitfalls before the review

Prove stage coverage first: intake plus processing equals total for every completed row, and the column totals satisfy 10 + 24 = 34 days. Then reconcile the completed count to the source export and keep excluded, canceled, or invalid cases visible as a separate check.

• Overlapping stages. Summing concurrent activities overstates elapsed time; the example assumes sequential stages.
• Mixing case types. A change in the share of complex work can move the average without any change in performance within a case type.
• Ignoring the unfinished tail. Completion-based summaries omit the old cases that have not finished.
• Changing the clock. Calendar days, staffed hours, and paused service clocks answer different questions.
• Turning blanks into zeros. Microsoft's MEDIAN documentation notes that referenced empty cells are ignored while zero values are included. A false zero can change the reported middle value.

Covirage is not affiliated with Microsoft. The process owner owns event definitions and action follow-up; the reporting owner checks timestamps and cohort counts. Compare the findings with deadline load per manager when the question becomes whether upcoming commitments fit available hours, rather than treating elapsed time as a capacity measure.

Review your timestamp export in Covirage

Bring a sample of your own case export with case identifiers, stage timestamps, status, and the review cutoff. Covirage maps the columns; its tools compute and check the durations and comparisons, and the external AI model explains the result with every figure traced to its rows. Create a free Covirage account, which takes about a minute, then set up your business and bring your file.

See self-service analytics for the supporting business use.