Executive Decision Queue Aging Review

Abstract

A review protocol for executive offices that need to measure how long decisions wait without turning the queue into a productivity score. It defines the population, separates submitter delay from decision delay, and tests whether aging is explained by missing evidence, unclear ownership, or insufficient decision capacity [1][2].

Methodology

Define the population as every request that entered the executive decision queue during a fixed quarter and required an approval, a choice among options, a ratification, or a direction. Exclude informational updates and requests decided by another leader. For each item, capture the first request date, the date the item became decision-ready, the decision date, the decision type, the owner, the preparer, the number of clarification cycles, and the final disposition. Compute age from the first request to the decision, and separately compute ready-to-decide time. Items still open at the cut-off are censored, not treated as completed. Report the count, the distribution, the share closed within the agreed service level, and the reasons for delay. Sample at least one item from each decision type and one that was closed without a decision. The protocol measures queue performance, not the quality of the underlying business choice [3][4]. Where the queue is distributed across email, chat, and meeting notes, describe the extraction method and the risk of missed items. Do not merge sources silently. If a request was made verbally and later confirmed in writing, use the written confirmation as the first request date and note the verbal origin. If the queue changed systems during the period, report the change and avoid comparing counts across incompatible definitions. The review team should include one person who did not work the queue, so the extraction rule is tested rather than assumed.

Population and measurement window

The first decision is what counts as a queue item. A request that asks the executive to approve a vendor, choose a candidate, ratify a committee recommendation, or give direction on a stalled project belongs in the population. A status update, a courtesy copy, or a request another leader can decide does not. Write the inclusion rule before extracting data [1].

Choose a window that is long enough to contain normal variation but short enough to reconstruct. A quarter is usually workable. Record the cut-off date and the state of every open item at that date. An item that is still waiting is not a completed decision, and counting it as one understates the backlog [2].

Separate the systems of record. Requests may arrive by email, chat, meeting, or a formal intake form. If the office has no single queue, the review should say so and describe how the sources were combined. A measurement built on the easiest source will overrepresent formal requests and miss the verbal ones that consume the most time [3].

Record the executive’s available decision capacity for the period, including travel, board commitments, and planned leave. Aging without capacity context invites the wrong conclusion. A queue that grows during a heavy travel quarter may be behaving normally; a queue that grows while capacity is stable is a different finding [1].

Timestamp discipline is the most common weakness. Email threads carry several dates, calendar invites are updated in place, and chat messages are edited. Choose the earliest verifiable request date and record where it came from. If the only evidence is a later reminder, say so and treat the age as a minimum rather than a precise figure. A queue review with honest uncertainty is more useful than a precise-looking number built on the wrong date.

Aging fields and service levels

Capture two dates for every item: when the decision was first requested and when the item became complete enough to decide. The gap between them is submitter delay. The gap from ready to decision is office delay. Combining them into one number hides where the problem sits [2].

Use a small number of age buckets, such as under three business days, four to ten, and more than ten. The cutoffs should follow the organization’s operating rhythm and the decision type. A routine approval and a major investment do not need the same service level [3].

Define the service level in writing before measuring. “Routine approvals are decided within two business days of becoming ready” is testable. “Decisions should be timely” is not. Record exceptions and the reason, rather than excluding them from the denominator [4].

Report the median and the oldest item, not only the average. A few very old items can be invisible in a mean and are often the ones that matter most. Pair the age measures with the count of open items so a shrinking queue is not confused with a faster one [1].

Report the measures by decision type as well as in total. A queue can meet its overall service level while routine approvals wait behind a small number of large investment decisions. Type-level reporting shows that trade-off. It also prevents a single complex item from distorting the picture for every simple approval in the same month.

Censoring, clarification, and attribution

Items still open at the cut-off are censored observations. Report them separately and state how long they have been open. Dropping them from the analysis makes the queue look faster than it is. Treating them as if they were decided today makes it look slower [2].

Count clarification cycles. A request that required three rounds of missing information is a different process failure from a request that was ready but waited for the executive. The clarification count also identifies intake templates that are not working [3].

Avoid attributing delay to a person without evidence. The queue may be slow because the approver is traveling, because the preparer did not supply the risk analysis, or because the decision depends on a committee that meets monthly. Record the dependency and the date it was resolved [1].

Check for re-opened decisions. An approval that is revisited after execution is not a queue delay, but it may indicate that the original brief omitted a consequence. Track re-opens separately and sample two or three to understand the cause [4].

External dependencies deserve their own field. A decision may wait on a customer, a regulator, a lawyer, or a committee that meets monthly. That waiting is not necessarily a queue defect, but it is still elapsed time the executive should see. Record the dependency, the expected date, and the date it cleared. Without that field, the office will be blamed for delay it did not control.

Defect categories and evidence quality

Classify each delayed item into one primary defect: missing authority or facts, unclear owner, unresolved consultation, capacity constraint, or disputed scope. A single primary category makes the results actionable; multiple tags make every item look like everything [2].

Test whether the brief met the office’s evidence standard. An approval without the amount, the executor, or the authority check is incomplete regardless of how long it waited. The review should separate speed from completeness, because a fast decision on a bad brief is not a success [4].

Sample a few decisions that were made quickly and later reversed. Speed is not quality. If the fast items show a higher re-open rate, the service level may be pushing the office to decide before the evidence is ready [3].

Record the source for each data field. A queue review assembled from memory and screenshots will not survive a challenge. Where the office cannot produce a reliable field, mark it as unavailable and limit the conclusion rather than filling the gap with an estimate [1].

If two reviewers classify the same sample, compare their categories before reporting. A disagreement usually means the category definitions are too broad. Resolve the definition, reclassify the sample, and record the change. A defect taxonomy that depends on who reviewed the item will not support a trend across quarters.

Review decision and limitations

At the review, the valid decisions are keep, revise, narrow, pause, or stop. “Keep” means the service levels and intake rules are working. “Revise” means changing the template, the owner, or the review cadence. “Narrow” means reducing the queue to the decisions that genuinely need the executive [4].

Assign each change an owner and a date. A review that produces a dashboard but no process change is an observation, not a control. Re-measure after one quarter using the same definitions so the comparison is meaningful [2].

State the limits. The population may be incomplete, the dates may come from different systems, and the decision types may not be comparable. The review does not measure whether the decisions were wise, only whether the queue moved and whether the evidence was present [1].

Keep the underlying item data under restricted access. A queue review contains the substance of pending business decisions, including personnel, legal, and financial matters. Aggregate reporting can be shared more widely than the item-level dataset [3].

Key findings

  • Queue age should be split into submitter delay and office delay; the combined number hides where the work is stuck [2].
  • Censored open items belong in the report, not outside the denominator [1].
  • Service levels must be defined by decision type before the quarter is measured [3].
  • Fast decisions with a high re-open rate are a quality signal, not a speed success [4].
  • The review is only a control if it produces an owned change and a re-measurement date [2].

Sources

  1. How CEOs Manage Time, Harvard Business Review
  2. Work Trend Index, Microsoft
  3. Work, Task and Project Management Resources, Asana
  4. Global Internal Audit Standards, Institute of Internal Auditors

Return to the Research Content library