A weekly review needs a few figures that describe the same process and period. Start with incoming work, completed work, the age of open assignments and the share of completed assignments returned for clarification. Give each measure an owner and a definition before building a dashboard. The example below is fictional and is not a report of an Estonian company's performance.
1. Define the boundary and the reporting cut-off
Choose one type of work, such as service assignments moving from an agreed request to an accepted result. Use the same assignment identifier through clarification, delivery and acceptance. The handover checklist sets out those transitions.
Write down the reporting period, time zone and cut-off time. Decide who maintains the records and who prepares the review. A task that arrives after the cut-off belongs to the next period, even if someone reads its message early. Store receipt, acceptance, return and cancellation events separately so changing the current state does not erase history.
Use the same scope each week. If a new team or task type is added, mark the change before comparing totals. An increased count after expanding coverage does not establish increased demand within the earlier boundary.
2. Keep a dictionary for the four measures
The dictionary is part of the report. A colleague should be able to reproduce a figure from its definition and source records.
Scroll sideways to see every column.
| Measure | Definition for this model | Source and owner | Missing-data rule |
|---|---|---|---|
| Incoming assignments | Distinct new IDs received during the period; exclude duplicates and reopened existing IDs | Receipt event; coordinator | Missing receipt time is an exception, not a confirmed count of zero |
| Completed assignments | Distinct IDs with recorded result acceptance during the period | Acceptance event; delivery coordinator | No acceptance record means completion is not established |
| Age of open work | Calendar days since receipt for IDs still open at the cut-off | Receipt time and final state; process owner | Show unknown ages separately; report how many records have a known age |
| Clarification-return share | Completed IDs with at least one return for clarification before acceptance, divided by all completed IDs in this period | Return and acceptance history; process owner | If the denominator is zero, report the share as not applicable |
Assign a named person to each source role in your own record. State when each figure was last refreshed. This model counts a completed ID once in the return numerator even if it was returned several times. Returns on unfinished work belong in the exception list, not in this completed-work denominator.
3. Reconcile incoming work with what remains open
In a fictional five-working-day review period, four assignments are open at the start. Six new assignments arrive. Five are completed and one is cancelled. Four remain open at the cut-off: 4 + 6 − 5 − 1 = 4.
Scroll sideways to see every column.
| Event | Fictional IDs | Count |
|---|---|---|
| Open at the start | A1, A2, A3, A4 | 4 |
| New during the period | N1, N2, N3, N4, N5, N6 | 6 |
| Completed | A1, A2, A3, N1, N2 | 5 |
| Cancelled | N3 | 1 |
| Open at the end | A4, N4, N5, N6 | 4 |
The incoming and completed counts need not match. Three of the completed assignments were already open at the start. An unchanged final backlog does not mean that no work moved through the process.
This reconciliation assumes no reopening, transfer or other entry or exit. If your real process has those events, add separate reconciliation rows. Do not force the totals to agree by deleting a task or treating cancellation as successful completion.
4. Read ages at one cut-off
Suppose the four open assignments in the same fictional example have known ages of 12, 5, 3 and 1 calendar days. The oldest is 12 days. The median is the mean of the two middle values after sorting: (3 + 5) / 2 = 4 days. All four records have a known age in this example.
Age is elapsed time, not staff effort. A task can spend much of its life waiting for a decision. Keep the reason for waiting beside the task so the review can identify a next action. Do not use a count of days as if it were the number of paid working days spent producing the result.
If a receipt date is missing, exclude that record from the age calculation and show the coverage explicitly, such as three known ages out of four open assignments. A low reported median with several unknown ages does not describe the entire open workload.
5. Keep the return denominator visible
In the example, only A2 among the five completed assignments was returned for clarification. It was returned twice before acceptance, but it is one affected assignment. The completed-work return share is therefore 1 / 5 × 100 = 20%.
Dividing by the six new assignments would mix two different populations. Counting both returns as two affected assignments would change the measure again. Either event counts or assignment counts can answer a question, but they need different names and denominators. This guide uses assignment counts.
Twenty percent is an illustrative result from five fictional completions, not a benchmark for a good team. The example does not establish a trend or an industry average. With no completed assignments, show the return share as not applicable rather than zero percent.
6. Separate waiting, work and data quality
A handover accepted the following day has an elapsed interval between events. That interval does not reveal how long someone actively worked on the assignment. Record active effort separately if you need it for a limited automation trial.
Before discussing a change in the figures, check duplicate IDs, missing timestamps, changed task types and late acceptance records. Keep corrections visible. An unexplained increase after cleaning the records is not automatically a deterioration in service.
The same discipline applies to external data. Our guide to reading Estonia's economic indicators explains why a label, unit and comparison period must travel with a figure. Internal workflow data need their own definitions; a national statistic cannot replace your missing task records.
7. Finish the review with owners and next actions
Bring the four figures, their coverage notes and a short exception list. Discuss which task needs an answer, which handover is incomplete and who will act next. Set a review date for unresolved exceptions. Avoid a dashboard that has a colour for every figure but no person responsible for the underlying record.
Use the first-month workflow plan to choose one limited change. Retain the baseline and keep the next period's definitions consistent. Targets should reflect your process and evidence; this model supplies no universal maximum backlog, ideal age or acceptable return rate.
8. Source and reusable record
- EIS: Digitalisation, checked 3 November 2026, provides the wider context of using data in business processes. The four measures and fictional ledger here are this publication's own working model, not an EIS reporting standard.
- For each measure record: name, definition, numerator, denominator, period, cut-off, source, owner, refresh time and treatment of missing values. Keep this dictionary with the weekly report.