Question and scope

This protocol measures the time between an operations item entering a shared queue and a named person accepting responsibility for its next action. The wait matters because an item can sit in plain sight without anyone treating it as their own work. The measure reports elapsed time and the share of items that wait beyond an agreed limit. It does not attempt to judge whether the original request was reasonable.

Scope is one queue and one defined operating period, such as a shared inbox across four consecutive weeks. Include items that entered during the window and items that were already open at the start. Exclude items that were withdrawn before anyone could act on them. State these boundaries in the report so a later reviewer can repeat the same count.

Population and unit of analysis

The unit is one tracked item, from its first recorded entry time to the moment a named owner accepts it. If a request contains several actions with different owners, split it into separate units when the system supports that. If the team tracked it as one item, keep one unit and note the lower precision. Use a dated export as the starting list rather than a recollection of the week.

Record the item reference, queue, entry time, acceptance time, current state, and the service level that applies. Include reopened items so the analysis can distinguish a first wait from a repeat wait. Exclude items that never carried a promised response time, but list the exclusion count. A clear unit definition is more valuable than a large sample with mixed meanings.

Definitions and measures

Acceptance means that a named person has taken responsibility for the next action, not that the item was merely opened. Record acceptance evidence such as a status change, an assignment, or a written reply from that person. Compute the median and 75th percentile wait for all accepted items. Report items still waiting at the end of the window separately, because their true wait is unknown.

Use fixed age bands such as under one hour, one to four hours, four to eight hours, one to three days, and over three days. Count each item once and show the counts behind each percentage. Split the result by queue, by shift, and by request type when the volume allows. Do not publish a single average that hides a slow corner of the process.

Evidence and data dictionary

Each item needs an entry timestamp, an acceptance timestamp, and a reference to the person who accepted it. A timestamp copied from a message is stronger than one reconstructed from memory. If the acceptance was verbal, record the date and the person who confirmed it. Keep the raw export unedited beside the coding sheet so the result can be rebuilt.

Define every field in a short data dictionary before coding begins. Include the source system, the meaning of each status, the owner field, and the rule for choosing between two timestamps. Mark unknown values instead of guessing them, and give each unknown value a short reason. A short dictionary that fits on one page prevents two reviewers from measuring different things under the same label.

Test design and sampling

Use a census when the queue is small enough to review completely and carefully. For a larger queue, draw a documented sample that covers every shift and every request type in the period. Freeze the sample before coding so the reviewer cannot choose items that support a preferred answer. Have a second person recode a subset and record disagreements before resolving them.

A practical starting window is four consecutive weeks or the most recent complete operating cycle. If the queue changes during the window, split the analysis at the change and report both parts separately in the final note. Do not compare a redesigned process with the period before it without saying so. The comparison is only honest when the definition of acceptance stayed stable.

Decision rules before the test

Write the decision rules before looking at the result. Decide what counts as acceptance, how long an item may wait, and which exclusions are allowed. State whether a documented delegate counts the same as the original owner. These choices change the number, so they must be fixed before anyone sees the data. Put the rules in the same file as the coding sheet.

Agree on the action that follows a long median wait. One option is to add an acceptance step that prevents an item from leaving intake without a named owner. Another is to review the oldest items each morning until the wait falls. Avoid starting with individual coaching, because the pattern may come from the intake design rather than from effort.

Analysis and reporting

Report the median wait, the 75th percentile, and the share of items waiting beyond the agreed limit. Show the oldest item, its queue, and the reason it waited. Pair the numbers with a short review of acceptance evidence, because a name in a field does not prove that the person agreed to act. The report should make the next management question obvious.

Break the result down by queue and shift, and show the item count behind each figure. A percentage based on five items deserves a different level of confidence than one based on fifty. If the sample is small, report counts and describe the uncertainty in plain language. Do not present an average as if it described every part of the queue.

Reporting template

Use a one-page template with four blocks: scope, median wait, share beyond the limit, and the action agreed at review. Keep the raw counts beside every percentage so a reader can see the size of the evidence. Add a short note for each exclusion and each process change during the window. The template should be short enough to complete during the review meeting.

Store the completed template in the same folder as the export and the coding sheet. A later reviewer should be able to reproduce the main number without asking anyone for extra context. If the folder holds several versions, mark the final one clearly and date the change. Evidence that cannot be found is treated as missing evidence whenever the result is challenged.

Limits and sources of error

The cited public sources frame records, internal control, and privacy in general terms; they do not supply a local result for a specific team. A fast acceptance does not prove that the work was correct, and a long wait does not prove that the team was careless. The measure also depends on how the team defines acceptance. A broad definition can count a glance as a transfer, while a narrow one can miss a real one.

Missing timestamps and reconstructed records are common sources of error. Exclude an item from the wait measure when either timestamp cannot be supported, but keep it in the queue count when the acceptance is known. Report the number of excluded items and the reasons beside the result. A clear, written exclusion rule is better than an exact number built on invented timestamps.

Review cadence

Review the result at a fixed date and end with one process change, one owner, and one check date. A repeated gap should change the intake rule or the acceptance step before it changes coaching. Keep the export, the coding sheet, and the notes together so the next reviewer can repeat the check. Compare like periods and state any process change beside the result.

Repeat the review after two full cycles, then move to a monthly rhythm if the measure is stable. Retire the measure if it stops informing a decision, because a metric nobody uses will be maintained badly. The purpose is a queue that the next person can act on, not a perfect record. Stop when the team can show that outcome consistently.

Sources