What rework looks like in a support desk

Abstract

Incomplete requests create work that is easy to feel and hard to count. An assistant rebuilds a document, rechecks a source, or restarts a meeting setup because a field was missing at intake. The rework is usually absorbed silently, so the desk looks slower without anyone seeing the cause. This review defines rework, sets a local measurement, and separates the cost of a bad request from the cost of a difficult task. [1][2][3]

Methodology

The review uses internal-control and records guidance to define a complete request and a repeat cycle. It then builds a local count from ticket, message, or task records using a fixed observation window and a short codebook. Two reviewers classify a sample so the definition does not drift. The method does not estimate a national rework rate and does not treat every follow-up as failure. [2][3][4]

The question and its boundary

The question is how often a request returns to the desk because information was missing, unclear, or contradictory. A return can happen before work starts, during production, or after review. Each stage has a different cost, and the review should count them separately. The boundary should exclude changes of mind that arrive with complete information. [1][3]

A request that changes because the business changed is not rework in this sense. It is a legitimate revision. The distinction matters because a team that counts every change as failure will stop accepting new information. The codebook should name the difference and give examples. Without that rule, the measure punishes good judgment. [2][4]

What counts as rework

Rework includes restarting a task, rebuilding a document, repeating a search, or renegotiating a meeting because the first request lacked a decision, a deadline, an audience, or access. It also includes time spent chasing the requester for a field that the form should have captured. Count the minutes and the number of returns, not only the existence of a problem. [1][3]

A return that takes two minutes to resolve is different from one that costs a day. The review should use bands, such as under fifteen minutes, fifteen to sixty minutes, and over an hour. The bands make the distribution visible without pretending to measure every second. They also help prioritize which missing field to fix first. [2][4]

Evidence from process and internal control

Internal-control guidance treats clear authorization, documentation, and review as part of a reliable process. A request that lacks an approver or a source record weakens all three. The guidance does not measure rework, but it explains why incomplete intake creates risk beyond lost time. The desk can use that framing when proposing a form or a triage rule. [1][2]

Records guidance adds a second reason to care. If the request cannot be reconstructed later, the team cannot show why a decision was made or who approved it. A short intake record becomes part of the audit trail. The review should count the requests that would fail that test, even if they did not cause a visible delay. [2][3]

A local measurement design

Choose a four-week window and list every request the desk handled. For each request, record the requester, the task type, whether the first submission was complete, the number of returns, and the total rework minutes. Use the same categories for every row. A simple sheet is better than a complex system that nobody maintains. [3][4]

Have a second reviewer classify one week. Compare the rework minutes and the reasons for each return. If the reviewers disagree, rewrite the example that caused the disagreement. Repeat the check after the rewrite. The goal is a definition that two people can apply the same way, not a perfect taxonomy. [2][4]

Common patterns

The most common gap is a missing decision or audience, not a missing file. A requester knows what they want to happen but not who needs to approve it or who will read it. The second most common gap is a deadline that ignores review time. Both gaps are cheap to fix at intake and expensive to fix later. [1][3]

A third pattern is access. The desk cannot open the source system and waits for a permission that nobody owns. A fourth is a request that mixes two outcomes, such as a board update and a client summary, in one message. The review should tag these patterns so the form and training can target them. [2][4]

Reporting and testing

Report the number of requests, the share that returned at least once, the median rework minutes, and the top three reasons. Add one example of a fix, such as adding an audience field to the form. Do not report a single percentage without the sample size and the window. A small desk may have too few requests for a stable rate. [3][4]

Test one change for a month. If the form now asks for the audience, compare the return rate for requests that used the new field with the rate for requests that did not. The comparison is not a controlled experiment, but it is better than a general impression. Keep the change if it reduces returns without adding friction. [2][4]

A worked example

Consider a hypothetical desk that handles sixty requests in four weeks. Twenty-two return at least once, and the coded sample shows that eleven are missing an audience or an approver. The median rework is twenty-five minutes, but four requests take more than two hours because they require a rebuilt deck or a new access request. The distribution matters more than the average. [3][4]

The team adds two fields to the intake form and rewrites the help text for the deadline field. In the next month, the return rate for requests that use the form falls, while requests that arrive by direct message remain unchanged. That split suggests the form helped, but it does not prove the training worked. The report should show both groups. [2][4]

What to test next

Test one field at a time. If the form now asks for the decision and the audience, compare returns for requests that complete both fields with returns for requests that complete only one. A difference is a useful signal, not a controlled result. Repeat the comparison for a second month before changing the whole form. [3][4]

If the return rate does not move, the problem may be the triage path rather than the form. A desk that accepts partial requests without sending them back teaches requesters that the fields are optional. The next test should change the response rule and measure whether the same gap persists. [2][4]

Form and training changes

A form is only one part of intake. The help text, the examples, and the first response from the desk all teach requesters what a complete request looks like. Update the examples when a new task type appears, such as a regulatory filing or a vendor renewal. A form with stale examples will collect stale information. [1][3]

Train new requesters in a short session rather than a long policy document. Show one complete request and one that returned, and explain the cost of the difference. The desk should also review its own triage language for clarity. A blunt return message can create defensiveness that a specific request for the missing field would avoid. [2][4]

Review the form after the first month rather than after the first quarter. Early feedback catches a confusing label while the examples are still fresh. Ask two frequent requesters to complete a sample request and watch where they pause. That observation is often more useful than a satisfaction survey. [1][4]

Limits and responsible use

This measure cannot prove that a requester is careless or that the desk is inefficient. It describes where information was missing and what it cost to recover. Some rework is healthy because it catches a bad plan before it reaches a client. The review should preserve that kind of check rather than eliminating every return. [1][3]

Use the result to improve the form, the training, and the triage path. Do not use it to rank individual requesters or assistants. A public league table will encourage people to hide problems rather than fix them. For related reading, see our guide to <a href="/blog/build-an-executive-support-intake-form-that-stops-vague-requests-at-the-door">building an intake form that stops vague requests</a> and our <a href="/services">executive support services</a>. [2][4]

Key findings

  • Count returns caused by missing decision, deadline, audience, access, or contradictory information, not every revision. [1][3]
  • Use time bands and return counts so a two-minute fix is not treated like a lost day. [2][4]
  • Internal-control and records guidance explain why a complete intake record matters beyond lost time. [1][2]
  • Classify a sample with two reviewers and resolve definition differences before reporting. [2][4]
  • Test one form or triage change for a month and compare return rates without ranking individuals. [3][4]

Sources

  1. Standards for Internal Control in the Federal Government, U.S. Government Accountability Office
  2. Security and Privacy Controls for Information Systems and Organizations, NIST
  3. Recordkeeping, Internal Revenue Service
  4. Stay Legally Compliant, U.S. Small Business Administration
  5. Hours of Work, U.S. Department of Labor

Return to the Research Content library