Work Sample Test

How work sample tests work

The employer picks a task that represents an important, frequent part of the job, gives every candidate the same instructions, materials and time limit, and scores the output with a rubric. The task can be done live (on site or over video), online in a timed environment, or as a take-home assignment.

Work samples overlap with a job simulation. The difference is mostly scale: a work sample is a single task; a simulation recreates a slice of the working environment, often with several tasks, interruptions and tools. Both sit under the broader heading of skills assessment.

Work sample test examples by role

• Customer service (email or chat): respond to three customer messages in 20 minutes using a short knowledge article. Score accuracy, tone and whether the customer's question was fully answered.
• Payroll or accounts payable clerk: reconcile a 40-line invoice batch against purchase orders and flag five planted errors.
• Recruiter: write outreach to two fictional candidates and screen five resumes against a job description, explaining choices.
• Maintenance technician: diagnose a fault on a training rig and talk through the fix.
• Data entry: key 50 records from scanned forms with a target accuracy and speed.
• Trainer: deliver a 10-minute segment of a real onboarding module to a small panel.

The pattern is the same: a real task, a fixed time, and scoring on what a good employee would produce.

Why they work

Work samples have long shown strong validity in selection research; in a 2022 re-analysis by Sackett and colleagues they remained among the better predictors, behind structured interviews. They also tend to show smaller group differences than cognitive tests, and candidates generally see them as fair because the link to the job is obvious. That face validity matters legally too: under the Uniform Guidelines, demonstrating that a test is a representative sample of the job (content validity) is one recognized way to show it is job-related.

Designing and scoring a fair task

1. Pick the right task. Ask the hiring manager which two or three tasks make or break performance in the first 90 days. Choose one that can be done in under an hour without company-specific knowledge.
2. Write clear instructions and give everyone the same materials, tools and time.
3. Build the rubric first. List three to five criteria with what a 1, 3 and 5 looks like. For a customer email: accuracy, completeness, tone, clarity and policy compliance.
4. Score blind where you can. Remove names from submissions before scoring.
5. Have two people score a sample of submissions and compare, to check the rubric produces consistent results.
6. Pilot it on two or three current employees to confirm timing and difficulty.

Keep a copy of the task, rubric and scores with the requisition. If a candidate challenges a decision, this documentation is your evidence.

Take-home assignments: length and drop-off

Long take-home tasks are one of the most common reasons strong candidates withdraw. People who already have a job, or are interviewing elsewhere, will not spend a weekend on an unpaid project. Our blog explains why candidates drop out over lengthy hiring assignments.

Practical guidelines:

• Keep take-home tasks to 60 to 90 minutes, and say so clearly.
• Put them late in the process, for a shortlist, not for every applicant.
• Do not ask for work you could use commercially, such as a real marketing plan or real code for your product. That invites complaints of free labor.
• Consider paying for longer tasks, especially for senior or contract roles.
• For volume roles, use a short, timed, online task of 10 to 20 minutes instead.

Common mistakes

• Testing skills the job does not require, like a complex spreadsheet for a role that uses a simple form.
• Scoring by gut instead of rubric, which brings back the bias the test was meant to remove.
• Different tasks or time limits for different candidates.
• Forgetting accommodations. Under the ADA, candidates may need extra time, assistive technology or a different format, provided the essential skill being tested is not changed.