Key Takeaways
- Choose a task that represents the actual role.
- Define observable scoring criteria before reviewing responses.
- A bounded work sample supplies one piece of evidence, not a hiring guarantee.
Assistant work sample design
An assistant candidate may describe their experience convincingly while a founder remains unsure how they would handle a real assignment. A work sample can make some of that performance visible, but only if the exercise represents the job and the evaluation method measures something relevant. An elaborate test can still answer the wrong question.
The central design decision is what the exercise is intended to establish. Are you checking a skill the person must already possess, observing how they apply a supplied procedure, or exploring how they ask for missing information? Those are different purposes and should not be collapsed into a single impression of whether the candidate “seems capable.”
Evidence reviewed
The US Office of Personnel Management's Work Samples and Simulations guidance describes exercises that resemble job activities. It notes that assessors may observe behavior or evaluate task results, and cautions that work samples are most appropriate when the relevant competencies are expected at entry rather than taught after selection. The page also discusses development and administration costs.
We checked this guidance on October 2, 2026. It is an authoritative assessment resource, but it does not validate a particular founder-created assistant exercise. Its discussion supports careful task alignment; it does not provide a success rate, hiring guarantee, or universal score threshold for the practical proposal below.
Review method and uncertainty
This article uses a focused reading of one assessment source, followed by an original example of exercise design. It is not a systematic review of selection research and does not report results from a candidate study. The example is intended to expose choices a founder should make before using a work sample.
Any selection process also needs to follow the organization's applicable employment, accessibility, compensation, and data-handling requirements. This article does not resolve those requirements. The person responsible for hiring should obtain the appropriate review instead of asking an assistant coordinating the exercise to make policy decisions by default.
A useful local exercise can produce evidence about a narrow task without establishing overall suitability for the role. Treat the result as one input alongside the rest of your process. Overstating what a short sample proves is a design failure even when the exercise itself is well run.
Start with a real job activity
Choose a recurring activity the assistant will actually perform. For an executive support role, this might involve organizing a meeting brief from supplied materials, identifying scheduling conflicts, or drafting a response that follows a stated permission boundary. Avoid tasks selected mainly because they are easy to grade.
Describe the activity before writing the test. Identify the inputs, expected output, available tools, and decisions the assistant is permitted to make. This job description becomes the basis for judging whether the exercise resembles the work or merely looks administratively complicated.
Separate essential knowledge from information the employer would normally provide. If the assistant will receive a calendar policy after starting, an exercise can supply that policy and assess its application. Penalizing a candidate for not guessing an undocumented preference would measure familiarity with the founder rather than readiness for the role.
Construct a bounded hypothetical packet
Consider an exercise in which the candidate prepares a meeting brief for a fictional founder. Provide a short calendar extract, an invitation, a participant list, and a few relevant background notes. Use invented business details so the exercise does not expose customer information or turn candidate work into an unapproved live deliverable.
Include a small number of deliberate ambiguities that resemble the job. For example, one participant's role may be unclear, or the invitation may imply a decision the founder has not authorized. The exercise should reveal whether the candidate notices and handles the gap, not whether they can solve a puzzle unrelated to ordinary work.
State which questions may be asked and how answers will be provided. If one candidate receives helpful clarification while another is told to proceed without questions, their results may not be comparable. Preserve the conditions under which the work was produced.
| Dimension | Observable evidence | What it does not prove |
|---|---|---|
| Information selection | Relevant facts retained and distractions omitted | Expertise in every future subject |
| Accuracy | Output matches supplied materials | Ability to verify unavailable sources |
| Judgment boundary | Unapproved commitments flagged | Authority to make those commitments |
| Communication | Reader can identify the next decision | General personality or cultural fit |
Write the scoring guide before reviewing work
Define the required elements and meaningful errors before seeing candidate responses. A meeting brief might need the purpose, participants, timing, open questions, and a clear distinction between facts and assumptions. The scoring guide should describe what acceptable performance looks like in each area.
Avoid criteria that merely reward your preferred style. A candidate may organize the brief differently while making every necessary decision visible. If layout matters to the job, explain why and provide the relevant standard rather than treating personal taste as an objective measure.
Separate serious errors from minor presentation issues. Inventing a confirmed commitment is different from using an awkward heading. Combining them into an undifferentiated error count can conceal the distinction between a correct but unpolished output and a polished output that changes the meaning of the source material.
Decide how to treat time
Speed can matter in some administrative tasks, but a time limit should reflect the exercise's purpose. A very short limit may emphasize familiarity with a tool rather than the judgment you intended to observe. A generous limit may still create an unreasonable burden if the task is unnecessarily large.
Record what the candidate was told about expected effort and available resources. If tool use is allowed, specify it consistently. Do not infer dishonesty from a method you never prohibited, or reward unsupported shortcuts because they produced a faster-looking answer.
Evaluate the result in relation to the conditions. A careful candidate who flags an unresolved question may have demonstrated better task judgment than someone who finishes quickly by guessing. The scoring guide should make that distinction visible before reviewers become attached to a polished result.
Calibrate reviewers with examples
Have the people reviewing the sample discuss a few hypothetical responses before evaluating candidates. Include an answer that is accurate but incomplete, one that is complete but makes an unauthorized assumption, and one that clearly identifies the missing information. This reveals whether reviewers interpret the criteria consistently.
When reviewers disagree, ask which observable feature produced the difference. “I liked this candidate's approach” is not enough. The discussion should return to the job activity and scoring guide, while allowing the guide itself to be improved if it misses an important dimension.
Do not revise the criteria selectively after seeing a preferred candidate's result. If the exercise needs a material change, consider how to preserve fairness and comparability across the process. The responsible hiring owner should decide how to handle that situation.
Use the follow-up conversation carefully
A follow-up can help explain the candidate's reasoning. Ask how they selected information, where they saw uncertainty, and what they would clarify before acting in a real setting. This can distinguish a deliberate choice from an accidental omission without requiring the reviewer to guess.
Keep the conversation connected to the work. It should not become an opportunity to reward candidates who confidently rationalize errors. Compare the explanation with the actual output and the conditions under which it was produced.
Also listen for problems in the exercise. If several candidates interpret a prompt differently, the instructions may be ambiguous. Candidate performance can reveal weaknesses in the assessment design as well as differences in skill.
A proportionate conclusion
A work sample is most useful when it makes a relevant, bounded behavior observable under clear conditions. The reviewed guidance supports that design principle, while the specific packet and scoring approach here remain a proposal to be evaluated in your own context. No single sample can guarantee a successful working relationship.
Before building an exercise, clarify the role through assistant support services. A free consultation can help separate work that needs existing skill from procedures you can reasonably teach during setup.
Frequently asked questions
Should we use a real task from the founder's inbox?
A fictionalized task is often easier to control and avoids exposing live information. Preserve the structure of the work while removing sensitive details. The hiring owner should decide how any candidate exercise is governed.
Is the highest score automatically the best hire?
No. The score reflects the defined exercise and its limitations. Interpret it alongside the rest of the selection process, and avoid treating a narrow task result as a complete assessment of the person.
What if candidates use different tools?
Decide whether tool choice is part of the job requirement. If it is not, focus on the relevant outcome and disclose allowed methods consistently. If a specific tool is essential at entry, make that expectation explicit before the exercise.
Sources
External sources cited in this article. Follow each link to review the original publisher and context.
