Measuring the Share of Sales Support Requests Reopened After Closure

A reopened request is a signal that the first closure did not hold. The request may have been marked done before the customer confirmed the fix, or the owner may have closed it because the queue was under pressure. Counting reopenings turns that vague frustration into a measurable pattern. This protocol defines the request population, the closure event, and the reopen event before any record is reviewed. It does not judge whether the original work was good or whether the customer is satisfied.

Research question and scope

The question is how often a sales support request is reopened after it has been marked closed, and where in the lifecycle the reopening occurs. The scope is one team's request queue during a fixed review window, not an industry benchmark. This matters because a closed request can look like completed work while the underlying task is still unresolved. The measure describes the frequency and timing of reopenings, not the quality of the original agent. Keep the observation window and the queue definition fixed so the result can be repeated.

Population and unit of analysis

Start with a dated export of every request that entered the queue during the review window. Record the request identifier, the requester, the account reference, the owner, the open date, the first close date, and any reopen date. The unit is one request, and a request that reopens twice contributes two reopen events but remains one request in the denominator. Include requests closed by the same person who opened them, since self-closure is part of the pattern. Exclude duplicates and test entries, and note every exclusion with a reason.

Set the observation window

Choose a window that is long enough to include normal variation but short enough to keep the records readable. Four to six weeks usually works for a steady queue, while a shorter window is acceptable for a new process with a small volume. Record the start and end dates in the analysis file and do not extend the window after seeing the first result. If the queue changes during the window, such as a new intake form or owner reassignment, note the change and consider splitting the result. A stable window is what makes the denominator meaningful.

Definitions and measures

A closure is recorded when the owner marks the request resolved in the approved queue and the record contains a completion note. A reopening occurs when the same request is returned to an open state, by the requester or by a reviewer, after that close. Do not count a new request that describes the same need as a reopening unless it is explicitly linked to the original. Compute the reopening rate as reopen events divided by closed requests, and report the rate separately for first closes and repeat closes. Also report the median days between close and reopen, because a same-day reopen differs from one that appears weeks later.

Why closure evidence matters

A close is only useful if the next person can see why the request was considered done. A completion note that says "handled" or "sent" does not show what was delivered, to whom, or when. When that evidence is missing, a reopening becomes the only way for the requester to reopen the conversation. The review should therefore code the presence of a completion note alongside the reopening event. A queue with many bare closes is likely to produce reopenings that look mysterious in the aggregate.

Test design

Use a census for a small queue and a documented sample for a large one. Freeze the export before coding so later edits cannot change the population. Have one reviewer code the closes and reopenings, and a second reviewer check a sample without seeing the first result. Record disagreements before resolving them, especially when a chat message implies a reopen that the system never logged. Where the record is ambiguous, mark it as unverified rather than assuming the request was reopened.

Evidence and data dictionary

Each row needs the request identifier, both dates, the owner at close, and the source of the reopen signal. Add the evidence type, such as a system state change, a requester message, or a reviewer note, and the date that evidence appeared. Keep customer names and commercial terms in the approved system, and use an internal reference in the analysis file. If the reopen was reported verbally, record who reported it and when the report was entered. The dictionary should be stable enough for a later reviewer to repeat the coding without asking the original analyst.

What a reopening can mean

A reopening can mean several different things, and the coding should separate them where the record allows. The customer may have found that the fix did not work, the owner may have closed the request before a dependent task finished, or a new person may have reopened it because the completion note was unclear. Each cause suggests a different repair. Do not collapse all reopenings into one quality score, because a single number will hide the specific process that needs attention.

Read the rate by request type

Reopenings rarely spread evenly across a queue. A billing correction may reopen because the system takes a day to reflect the change, while a data update may reopen because the owner misunderstood the field. Group the rate by request type and by source, then compare the groups with enough volume to be meaningful. Do not publish a rate for a group with only a handful of requests, because one event will look like a trend. The useful output is a short list of request types that deserve a closer review.

Where reopenings cluster

Look for reopenings that cluster around a handoff, a system change, or a particular requester. A cluster after a representative change suggests that the completion note did not carry enough context. A cluster after a tool update may point to a timing problem rather than a human error. Group the events by week and by owner to see whether the pattern is stable or isolated. A cluster is a prompt for a closer look, not proof of a cause.

Limitations and next action

The result can support a change to the close rule, the completion note, or the reviewer who approves a closure. It cannot prove that the customer is satisfied or that the original work was wrong. A low reopening rate can hide silent failures if requesters have stopped reporting problems. Pair the measure with a small sample of requester confirmations and a review of completions that contain no evidence. Report reopened, unverified, and duplicate-linked events separately so the repair can match the gap, and end with one change, one owner, and one recheck date.

Keep the measure separate from performance review

The reopening rate is a process signal, not a reliable measure of individual performance. An owner who handles complex requests may see more reopenings than a colleague with simpler work. Using the number in a performance review will encourage people to avoid difficult requests or to argue about definitions. Keep the result at the queue and process level, and use a separate review for individual coaching. That separation protects the honesty of the coding and keeps the focus on the repair.

Report the result

Report the number of closed requests, the number with at least one reopening, and the total reopen events. Show the rate, the median days to reopen, and the share of closes that lacked a completion note. Add a short example of a supported reopen and one that was coded as unverified, with identifying details removed. State the limitation beside the numbers rather than in a footnote. The reader should be able to see what was measured and what the team plans to change.

Review cadence

Run the first review after a full month of queue data and repeat it whenever the close rule or queue tool changes. A monthly rhythm fits a high-volume support queue, while a quarterly check may be enough for a small team. Keep the same window length so the rates remain comparable. Archive the export, the coding sheet, and the notes beside the finding. The cadence should be short enough to finish before the pattern goes stale.

Sources

The sources below provide general context for recordkeeping and internal control. They do not supply a local reopening result or a benchmark for this team, and they were checked on November 6, 2026. Use them for background only, and keep the local observation separate from any external claim. The source list does not replace the queue record or the owner review.