Queue Age Anomalies in Outsourced Back-Office Work

Research findings

This protocol asks how often the age shown in an outsourced work queue disagrees with the event history of the underlying item. Define the population as items open at any point during a declared observation window in one workflow. The unit is one item linked to its creation event, state changes, owner transitions, and close event. Freeze the window, systems, and inclusion rules before viewing results.

Age anomalies matter because a queue can look healthy while items are older than reported. A reset timestamp, a reopened case, or a migrated record can make an aging item appear new. The study describes the frequency and shape of those disagreements; it does not rank a team or prove that a specific cause produced a specific delay.

Build the frame from the queue export and the audit history, not from the current status field alone. Include open, closed, reopened, cancelled, and transferred items so the denominator is explicit. Record every exclusion with its reason. If an item appears in two queues, define whether it is one unit or two before counting, and keep that rule stable.

For each item, capture the created time, the first owner assignment, every state change, the current owner, and the close time. Preserve the timestamp each system displays and note the time zone. Where a migration changed the identifier, keep a crosswalk. Do not replace an earlier timestamp with a later one because the later value looks more consistent.

Define an age anomaly before measuring it. Examples include a current age shorter than the time since the last owner change, a close time earlier than the last update, or a reopened item whose age restarted at zero. Record which rule flagged the item, the conflicting fields, and the reviewer who confirmed the conflict. A rule that depends on private knowledge is not reproducible.

Discussion

Report the eligible count, the observed count, the number with complete event history, and the number that could not be classified. Present anomalies by workflow, owner group, age band, and reopen status. Show the share of items whose age could not be verified separately from the share whose age was verified and found inconsistent.

Build a chronology for a protected sample of anomalies from creation through each state change and close. Separate waiting on the company owner, waiting on a customer or vendor, system delay, and time outside the service window. If the queue records only the current state, the chronology may be incomplete; say so rather than inferring missing intervals.

Implications and limits

Use a second reviewer on the same protected subset with the same definitions and event export. Compare the classification of each anomaly and the treatment of missing timestamps. Disagreements about what counts as an anomaly return to the owner who can clarify the workflow rule. Do not resolve them by majority vote.

Protect the extract with stable study keys, named access, and approved storage. Keep the identity crosswalk in a separate restricted location. NIST and CISA sources inform access and logging practice, while GAO guidance supports examination of provenance and completeness. None of them validates this queue or certifies a provider.

A queue-age study is descriptive. Results depend on the event history the systems retained, the observation window, the workflow mix, and the reviewer's classification. Test the findings against alternative windows and against a sample that includes records with missing history. If the conclusion changes, report both readings and name the decision that depends on the definition.