Assistant Time-to-Competence: A Measurement Design

Table of contents

• Question and method
• Define competence levels
• Choose task families
• The start event
• The end event and evidence
• Open cases and censoring
• A reporting table
• Pair with quality
• Limits
• Sources
• Frequently Asked Questions

Question and method

Research question: How can an executive-support team measure how long it takes a new assistant to reach independent competence, without reducing onboarding to a single number or rewarding a weak quality bar?

This paper defines a measurement design. It does not report an observed onboarding duration or claim a benchmark. It focuses on the end event, the task families, the evidence of competence, and the handling of assistants who have not yet reached the threshold.

Onboarding is often described in weeks: "It takes about a month." That statement mixes training, observation, and independent work. A task-level measure is more useful because different tasks become independent at different rates.

Define competence levels

Use three levels per task family:

1. Observed: the assistant has watched the task and can describe the steps.
2. Assisted: the assistant performs the task with a reviewer checking the output before it leaves the team.
3. Independent: the assistant performs the task without pre-delivery review, within the written authority and quality bar, for a defined number of consecutive runs.

The end event for time-to-competence is the first independent run that meets the quality bar and is accepted without material rework. A run that is independent only because nobody checked is not evidence.

The level definitions should be written for each task family. "Can manage the inbox" is not a level; "can triage 40 messages into the three lanes with no misrouted escalation for two consecutive days" is.

Choose task families

Group tasks by the skill and risk involved:

• calendar coordination
• inbox triage and drafting
• meeting notes and action tracking
• travel coordination
• vendor follow-up
• recurring reporting
• document preparation
• spend administration
• sensitive or personal support

Measure each family separately. An assistant may reach independence on calendar work in two weeks and on reporting in six. Averaging them hides the difference and gives a misleading onboarding time.

The start event

Start the clock when the assistant begins the task family, not on the first day of employment. If calendar training starts in week two, the calendar measure starts there. Record the employment start date separately for context.

Also record prior experience. An assistant with five years of calendar coordination may reach independence faster than someone new to the tool. The measure describes this onboarding, not the person's general capability.

The end event and evidence

The end event needs evidence:

• the task family and level
• the date of the first independent run
• the quality check applied
• the reviewer
• the number of consecutive accepted runs
• any authority limits that remain

If the task never becomes fully independent because of its risk, define a "supported independence" level with a sample review. Do not force every task into the same end state.

Open cases and censoring

Some assistants will still be in onboarding at the reporting cut-off. Report their current level and age per task family. Do not assign them a completion time or exclude them silently.

This is the same censoring issue that affects time-to-fill and other duration measures. Dropping open cases makes the completed cases look faster. Treating them as very slow makes the average meaningless. Report age and level.

A reporting table

The numbers are illustrative structure, not findings. An organization must populate the table from its own records and state the period.

Pair with quality

A short time-to-competence is not automatically good. Pair it with:

• rework rate after independence
• post-acceptance corrections
• escalation accuracy
• the share of independent runs sampled for review

If independence arrives quickly but rework rises, the quality bar was too low or the review was too shallow. The measure should trigger that question, not answer it.

Limits

The design assumes the team can identify task families and evidence of independent completion. In a small team, the owner may be the only reviewer, which creates a conflict of interest and a capacity constraint. Record who reviewed and when.

It also does not measure the assistant's experience of onboarding, the quality of training, or retention. Those require separate feedback and should not be inferred from speed.

The O*NET executive administrative assistant profile describes the task domain but not a training standard. The U.S. Department of Labor hiring guidance and Bureau of Labor Statistics provide context. The NIST Privacy Framework is relevant if onboarding records include personal data about the assistant.

Sources

1. O*NET OnLine: Executive Administrative Assistants
2. U.S. Department of Labor: Hiring
3. Bureau of Labor Statistics: Occupational Outlook Handbook
4. NIST Privacy Framework
5. U.S. Office of Personnel Management: Delegation

Frequently Asked Questions

Is time-to-competence the same as time-to-productivity?

Not necessarily. Competence is demonstrated independent work at the quality bar. Productivity also depends on volume, context, and the rest of the team. Measure them separately.

How many consecutive runs count as evidence?

Set the number by task risk. Three consecutive accepted runs may be enough for low-risk calendar work; high-risk spend work may need a longer sample and a second reviewer.

What if an assistant never reaches independence on a task?

That is a finding, not a failure of the measure. Review the task design, the training, and the role fit. Some tasks may need to be reassigned or supported permanently.

Review cadence

Review the measure at 30, 60, and 90 days with the assistant and the manager. Discuss which tasks reached independence, which still need review, and whether the support provided matched the task risk. Do not use the milestone as a performance rating. Use it to decide the next task and the next review date.

A good milestone review includes the evidence the assistant produced, not only the manager’s impression. Use a small sample of completed tasks and check whether the brief, the source, and the approval were correct without a second pass. If the evidence is thin, extend the milestone rather than declaring readiness. The point is a reliable next step, not a faster calendar.

Record the date of each milestone review and the specific evidence used. If the review is delayed, note why and whether the delay changed the decision. A milestone that is never reviewed is only a start date with a label. Keep the evidence with the milestone note so the next review can compare like with like. The note also shows the assistant what good evidence looks like in this role. That habit turns a one-time milestone into a durable onboarding record.