How often does an assistant task brief need a follow-up question?
A proposed protocol for measuring whether a task brief lets the assistant start without clarification.
Headline statistic
9 named sources and 8 protocol sections.
Methodology: A desk review of public guidance translated into a repeatable measurement protocol for task support.
Key stats
- 9 public sources reviewed
- 8 protocol sections
- November 9, 2026 verification date
Key takeaways
- Define the unit and the denominator before collecting evidence.
- Report missing and unavailable evidence separately.
- Keep the owner and the evidence date beside the finding.
Question and scope
This is a proposed research protocol, not a report of completed findings. It asks how often a task brief needs a follow-up question. The result would describe the brief, not the assistant. A control that depends on memory is not yet a control.
The protocol does not measure whether the task was valuable. It measures whether the brief was usable. Record the date and the source so the next reviewer can repeat the check. Review the first few notes together before the process expands.
| Item | Finding or control | Source |
|---|---|---|
| Unit | task support | Method note |
| Evidence | Recorded source | Public guidance |
Population and unit of analysis
Define the population as task briefs issued during a stated period. The unit is one brief, with the outcome, source, boundary, and follow-up questions recorded. Use the same format for every handover so the next reader can scan it. Keep the evidence with the record rather than in a separate log.
Link the brief to the task so the same brief is not counted twice. Separate what was observed from what was assumed. Ask whether a new reviewer could act on the record without a conversation. Name the owner in the same line as the action.
| Item | Finding or control | Source |
|---|---|---|
| Unit | task support | Method note |
| Evidence | Recorded source | Public guidance |
Definitions and measures
Define usability as the absence of a clarifying question before the first draft. Report the follow-up rate and the reasons. Keep the evidence with the record rather than in a separate log. A short note with four fields is easier to use than a long narrative.
Report the reasons separately to show which field is missing. A short note with four fields is easier to use than a long narrative. Keep the evidence with the record rather than in a separate log. Record the exception and the person who approved it.
| Item | Finding or control | Source |
|---|---|---|
| Unit | task support | Method note |
| Evidence | Recorded source | Public guidance |
Evidence standard
Use the brief and the message log. A question without a time is weak. Separate what was observed from what was assumed. Keep the boundary visible in the place where the work is assigned.
Record the date and the reviewer. Keep the boundary visible in the place where the work is assigned. Keep the evidence with the record rather than in a separate log. Test the rule on one real request before applying it to the queue.
| Item | Finding or control | Source |
|---|---|---|
| Unit | task support | Method note |
| Evidence | Recorded source | Public guidance |
Sampling and collection
Use a census of the briefs. If sampling is necessary, stratify by task type. Name the owner in the same line as the action. Record the date and the source so the next reviewer can repeat the check.
Pilot the classification on a few briefs and compare the reviewers. Clarify what counts as a clarifying question. Name the owner in the same line as the action. Test the rule on one real request before applying it to the queue.
| Item | Finding or control | Source |
|---|---|---|
| Unit | task support | Method note |
| Evidence | Recorded source | Public guidance |
Analysis and reporting
Report the follow-up rate by task type. Show the fields that were missing most often. Present the results as a description of the brief, not a judgment about the assistant. Review the first few notes together before the process expands.
| Item | Finding or control | Source |
|---|---|---|
| Unit | task support | Method note |
| Evidence | Recorded source | Public guidance |
Limits and sources of error
The assistant may ask a question out of caution, not because the brief is weak. The brief may be discussed before it is written. Test the rule on one real request before applying it to the queue. Keep the boundary visible in the place where the work is assigned.
| Item | Finding or control | Source |
|---|---|---|
| Unit | task support | Method note |
| Evidence | Recorded source | Public guidance |
Review cadence
Repeat the review after a template change or at the same interval each quarter. Assign the follow-up to the operations lead and set a date. Name the owner in the same line as the action. Ask whether a new reviewer could act on the record without a conversation.
| Item | Finding or control | Source |
|---|---|---|
| Unit | task support | Method note |
| Evidence | Recorded source | Public guidance |
Questions people ask
Sources
- U.S. Small Business Administration: Public guidance consulted for general context.
- GAO Green Book: Public guidance consulted for general context.
- NIST Privacy Framework: Public guidance consulted for general context.