How to choose real work for a hiring exercise

To choose real work for a hiring exercise, look for a bounded responsibility that matters in the job, can be approached with information you are prepared to provide, and produces actions or outputs a reviewer can inspect. Remove production access and unnecessary sensitive information. Decide how the work will be reviewed before inviting applicants.

That sounds simple. The difficult part is deciding what to preserve from real work and what to change.

Begin with a responsibility, not an interesting file

A memorable customer conversation, project document, or incident record may look like a good exercise. But the source material is not the exercise. It is evidence about a responsibility.

Start by naming that responsibility in one sentence:

• Investigate a delayed shipment and decide what to tell the customer.
• Review a draft analysis and identify what must be resolved before publication.
• Plan a handover when a dependency will not be completed before the end of a shift.

The sentence should contain something the person in the role genuinely needs to do. OPM describes work samples as tasks or activities that mirror work performed on the job. It also cautions against using a work sample to test competencies that the employer intends to teach after selection (OPM, Work Samples and Simulations).

This distinction prevents a common error: selecting material because it is dramatic rather than because it represents the work.

The work selection checklist

This checklist is an editorial tool for designing a useful exercise. It is not a validated assessment framework and does not replace professional or legal review.

Job relevance

• The responsibility matters in the role. A good or poor decision would have a meaningful effect on customers, colleagues, quality, cost, risk, or progress.
• The exercise reflects work expected at entry. It does not primarily test a procedure, product detail, or internal shortcut that the company normally teaches after hiring.
• The slice is representative enough to discuss. It reveals something relevant about how the person investigates, communicates, decides, executes, or hands work over.
• The difficulty is appropriate for the level. An entry-level applicant is not being judged against the context and authority of an experienced employee.

Candidate conditions

• The objective is understandable. The applicant can tell what responsibility they have accepted and what a useful outcome would look like.
• Necessary context can be supplied. The company can provide relevant instructions, definitions, records, and constraints without giving away the judgment being explored.
• Unknowns are deliberate. If information is missing, the exercise makes it possible to ask, record an assumption, pause, or escalate. Missing information is not an accidental trap.
• The effort is proportionate. The work can be completed or responsibly handed over within the time and effort communicated to applicants.

Inspectable evidence

• The work produces evidence. Reviewers can inspect decisions, messages, notes, outputs, changes, or outstanding obligations—not just a final “correct” result.
• More than one defensible path is allowed. Reviewers know which constraints are firm and where judgment can reasonably differ.
• A review guide can be written in advance. The company can describe what reviewers should notice, what evidence supports it, and which conclusions require follow-up.
• The same core conditions can be maintained. Applicants receive comparable instructions and reviewers use a consistent starting set of questions.

Boundaries and information

• The exercise can run outside production. Applicants do not need access to live customer, employee, financial, clinical, or operational systems.
• Unnecessary sensitive information is excluded. Names and obvious identifiers are not the only concern; rare combinations, free text, attachments, and operational details can also identify people or expose the company.
• De-identification receives human review. The team does not assume that replacing names makes arbitrary source material safe.
• The material is approved for applicant use. Someone accountable has reviewed the complete environment, not only the initial file.

Hiring process

• Applicants know what to expect. The invitation explains the task, expected effort, resources, timing, review, and route for questions or accommodations.
• The employer knows how the evidence will be used. The exercise has a defined place alongside interviews and other selection information.
• Reviewers can separate observation from interpretation. They can state what happened before deciding what it might mean.
• The company can maintain the exercise. The content will be reviewed when the job, systems, policies, or applicant population changes.

If several boxes remain uncertain, do not solve the problem by adding more content. Narrow the responsibility first.

An illustrative example: a delayed shipment

The following example is fabricated. It is not customer research or a description of a currently available Calavion environment.

The historical material

A support team has a conversation about a delayed order. The employee apologised, contacted the warehouse, offered a replacement, and closed the ticket after the customer accepted.

Copying that conversation would create a weak exercise. The applicant could be rewarded for reproducing one sequence without understanding why it was reasonable. The historical employee may have known an unwritten escalation rule. The carrier’s status, replacement stock, customer history, and cost threshold may also have changed.

The responsibility beneath it

Investigate a delayed shipment, communicate with the customer, and leave the order in a responsible state.

Context to preserve or recreate

• The customer’s stated need and previous messages.
• The order, carrier, and warehouse information available at the start.
• Relevant service commitments and the applicant’s authority.
• The routes available for checking information or requesting help.
• A later update that responds to an action or introduces a realistic change.

Material to exclude or replace

• Real names, addresses, order numbers, contact details, and free-text identifiers.
• Internal comments that reveal the historical outcome as the expected path.
• Commercially sensitive thresholds that are unnecessary for the responsibility.
• Company-specific abbreviations that are not explained.
• Systems or records unrelated to the decision.

Evidence worth discussing

• Which sources the applicant checked before responding.
• What the applicant told the customer and what remained uncertain.
• Whether the chosen remedy was within the stated authority.
• How the applicant handled the warehouse dependency.
• What remained open, who owned it, and what should happen next.

The exercise is now about a responsibility. It is not a memory test for the employee’s sequence.

Why the historical outcome is not an answer key

Past work is valuable because it reveals context, constraints, dependencies, failure modes, and consequences. It does not automatically establish the only correct response.

There are at least five reasons:

1. The employee had context that may not appear in the record. A copied answer may reward access to hidden knowledge. 2. The conditions may have changed. Policies, inventory, tools, authority, and risks move over time. 3. The outcome may reflect habit rather than best judgment. “This is what happened” is not the same as “this is what must happen.” 4. Several approaches may satisfy the same constraints. A thoughtful review should preserve defensible variation. 5. A final outcome hides the route. Two applicants can reach the same result after very different investigations, messages, and risk decisions.

Use historical work to create three review layers instead:

• Firm constraints: actions that were outside authority, required information that could not be ignored, or obligations that had to remain visible.
• Observable evidence: checks performed, messages sent, decisions made, outputs created, and work left pending.
• Conversation prompts: questions that explore trade-offs, alternatives, and reasoning without pretending to read the applicant’s mind.

OPM recommends linking writing assessments to critical job competencies and using standardized review and scoring procedures (OPM, Assessment Methods). In the United States, the EEOC also emphasizes that selection procedures should be job-related and suitable for the purpose for which they are used (EEOC, Employment Tests and Selection Procedures). Employers should obtain advice appropriate to their jurisdiction and use case.

What should be excluded from a hiring exercise?

Exclude work when its main value depends on any of the following:

• Undisclosed insider knowledge.
• Access to live production systems.
• Personal or sensitive information that is not necessary.
• A volume of unpaid work disproportionate to the hiring decision.
• A deliverable the company intends to use as free production work.
• A single hidden answer that reviewers cannot justify from stated constraints.
• Skills the employer expects to teach after hiring.
• Conditions that reviewers cannot reproduce or compare responsibly.

De-identification can reduce privacy risk, but NIST notes that some de-identified information can be re-identified (NIST IR 8053). Treat data reduction, replacement, fabrication, and company review as deliberate choices—not as a promise that a transformation is perfect.

Decide how you will discuss the work before you run it

For each important moment, write three lines:

1. What can a reviewer observe? 2. What are two plausible interpretations? 3. What neutral question would help distinguish them?

Example:

• Observation: The applicant waited for the warehouse response before offering a replacement.
• Possible interpretations: They protected against unnecessary cost; or they allowed avoidable customer uncertainty.
• Question: “What information were you waiting for, and what would have caused you to act sooner?”

This is the bridge from a work exercise to a useful interview.

Where Calavion fits

In Calavion, a company submits selected work, Calavion prepares a recreated interactive environment with the company, and the company approves the version applicants work through. Managers receive an evidence-linked review and can return to specific moments through Replay.

The product does not make every source file suitable, turn a historical response into truth, or remove the employer’s responsibility to review the exercise and interpret the evidence.