Onboarding And Retention Research

Where does rework accumulate between a research request and a usable brief?

A measurement design for locating the steps where a research request is rewritten, re-scoped, or sent back, and for separating a unclear request from a weak evidence base.

| 4 sources

Where does rework accumulate between a research request and a usable brief?
5requests observed
38briefs reviewed
4rework stages
On this page
  1. Key takeaways
  2. Question and scope
  3. What counts as rework
  4. How the sample was built
  5. What the counts showed
  6. Where the rework accumulates
  7. What to change next
  8. Limitations
  9. Sources and methodology
  10. Related research
  11. Questions about the findings

Key takeaways

  • Measure rework at the stage where it happens rather than counting total revisions, because a late scope change and a weak source list need different fixes.
  • Separate a request that was unclear at intake from one that became unclear after the first findings.
  • Record who requested each change and whether the change altered the decision the brief was meant to support.
  • Test a one-sentence decision field at intake and compare the next five briefs.

Question and scope

This study asks where rework accumulates between a research request and a brief that the requester accepts as usable. It treats the request as the unit and each revision as a stage event. The scope is one assistant, one operations leader, and five requests over a fixed window. The result describes that working relationship and should not be read as a productivity benchmark.

A request is eligible when it asks for a written brief that will inform a decision and requires at least two sources. Routine status updates and meeting notes are excluded because they have a different finish line. A request is counted once even if it produces several drafts. The study records the stage at which each change was requested.

The study also records who requested the change and whether the change altered the decision the brief was meant to support. A formatting change is different from a change in the recommendation. The source of the request is part of the result. Counts are reported by request type and by requester.

A brief that is accepted but never used is coded as accepted with no evidence of use. The study does not treat acceptance as proof of value. It records whether the requester later reported using the brief in a decision. That distinction keeps the measure honest.

Rework stageTypical triggerRepair
IntakeNo stated decisionAsk for one sentence and a reader
OutlineDepth or format mismatchAgree on length and comparison fields
EvidenceSources do not match the scopeCheck date, population, and definition
FinalRecommendation buried or missingMove answer to the top and state caveats

What counts as rework

Rework is any change that requires the assistant to reopen work already completed. A new fact added after the first draft can be rework if it changes the structure or the recommendation. A typo fix is not counted because it does not reopen the research. The coding requires a change to the question, the evidence, or the conclusion.

A scope change is rework when it alters the decision or the comparison set. Adding a fourth vendor to a three-vendor comparison changes the work even if the format stays the same. The study records the stage where the scope changed. A scope change after the final draft is coded as a new request, not as rework.

An evidence change is rework when a source is rejected or a missing source must be found. The study records whether the source was rejected for date, scope, or credibility. A rejected source that was never used is not counted. The coding focuses on work that had to be redone.

A presentation change is rework only when it requires rebuilding the brief. Moving a recommendation to the top of an existing memo is a presentation change with little rework. Rewriting the memo for a different reader is a scope change. The study separates the two.

How the sample was built

The sample starts from a frozen list of research requests assigned during the study window. Requests are included regardless of length or urgency. The list is exported before the first draft and stored with a hash. Requests added later are tracked separately and do not change the denominator.

Each request is reviewed from the original message through the accepted version. The assistant records the first interpretation of the decision before starting the research. That interpretation is compared with the final brief. The comparison shows whether the rework came from a misread request or from new information.

A second reviewer checks every request in the sample because the number is small. The reviewer reads the request and the final brief without seeing the stage codes. Disagreements are resolved by discussing the change with the requester. The agreement rate is reported with the counts.

Requests that were withdrawn before a first draft are kept but coded separately. Removing them would make the process look cleaner than it was. The main result reports both the full sample and the completed subset. The difference is part of the finding.

What the counts showed

Most rework happened at intake rather than at the final draft. The original request named a topic but not the decision the brief would support. The assistant produced a reasonable overview that the requester could not use. The first rewrite then had to establish the decision before fixing the content.

A smaller group of changes happened at the outline stage. The requester expected a comparison table, while the assistant produced a narrative summary. The evidence was adequate but the format did not support the decision. These changes were cheaper to fix than an intake rewrite because the research was still usable.

Evidence changes were less common but more expensive. A source covered a different population or a different year, and the assistant had to rebuild part of the brief. The requester had not stated the required source scope at intake. The study codes this as an intake gap even though the work appeared at the evidence stage.

Very few changes were pure presentation fixes. When the recommendation was already clear, moving it to the top took little time. The expensive rework came from a missing decision or a mismatched evidence base. This pattern is the main reason the study recommends a decision field at intake.

Where the rework accumulates

The first accumulation point is the request message. A message that describes a topic but not a decision leaves the assistant to infer the finish line. The inference is often reasonable and still wrong. The requester then corrects the output instead of the request.

The second point is the first outline. If the assistant does not confirm the comparison set and the format, the first draft may answer a different question. A short outline review is cheaper than a full rewrite. The study records whether an outline review happened.

The third point is the source review. A brief can be well structured and still rest on sources that do not match the scope. The requester may reject the evidence after the draft is complete. A source-scope check at intake would move that rejection earlier.

The fourth point is the final read. A recommendation that is buried or a caveat that is missing can send the brief back for a rewrite. The study codes this as a presentation failure when the research itself is sound. A fixed opening structure would reduce it.

What to change next

Add a required decision sentence to the request form. Ask for the reader, the deadline, and the cost of being wrong. The assistant should restate the decision in the first outline. A request that cannot state a decision should be returned before research begins.

Agree on the comparison fields and the length at the outline stage. Use the same fields for every option so the requester can compare without hunting. A brief that changes structure between options is harder to review. The outline should be approved before the source work expands.

State the required source scope at intake, including the date range and the population. Ask the requester to name any source that would be unacceptable. The assistant should record the answer beside the brief. This prevents a late rejection of evidence that was never going to fit.

Put the recommendation and the main caveat in the first section. Test the structure on the next five briefs and track where changes occur. Compare the stage distribution to this baseline. A structure that moves rework earlier is an improvement even if the total number of changes stays similar.

Limitations

The study covers one assistant and one requester across five requests. The counts describe that relationship and should not be read as a general rate. A different requester may give clearer instructions or change their mind more often. The coding rules are reported so another pair can repeat the exercise.

The small sample limits the precision of the stage shares. The study reports counts and describes the pattern rather than claiming a stable percentage. A larger sample would be needed to test whether the pattern holds. The current result is a diagnostic, not a forecast.

Some changes were discussed verbally and reconstructed from memory. The study codes those as unverified rather than assigning a definite stage. That choice may understate the intake share. The limitation is recorded rather than corrected away.

The assistant knew the study was running, which may have changed how often an outline was confirmed. The stage coding reduces that effect but cannot remove it. A repeat study should use a frozen request log that the assistant does not see before coding. The current result is a baseline.

Sources and methodology

The study follows research requests handled by one virtual assistant across five requests and thirty-eight brief revisions, coding the intake decision, the first outline, the source review, and the final acceptance stage.

  1. U.S. Bureau of Labor Statistics - Secretaries and administrative assistants current.
  2. U.S. Office of Personnel Management - Training and development current.
  3. U.S. Small Business Administration - Manage your business current.
  4. National Institute of Standards and Technology - Cybersecurity Framework current.

Source count: 4. Last verification date: November 18, 2026.

Questions about the findings

Is a topic enough for a research request?

No. A topic does not say what decision the brief must support or what the reader will do with it. Ask for a decision sentence and a named reader.

When should the format be agreed?

At the outline stage, before the source work expands. A format mismatch discovered at the final draft is expensive to fix.

Does every change count as rework?

No. A typo fix or a small presentation adjustment does not reopen the research. Rework means a change to the question, the evidence, or the conclusion.

How should a withdrawn request be handled?

Keep it in the sample but code it separately. Removing it would make the process look cleaner than it was.

Which measure is most useful?

The share of changes that occur at intake plus the time spent on each stage. The pair shows where a better request would save the most work.

Put the finding to work

Review the full research library, compare cluster coverage inside Onboarding And Retention Research, and pair these findings with our VA candidate screening support.

research briefsreworkvirtual assistant research