work-sample-tests

# Work Sample Tests ## Why It Matters Work samples are among the most predictive selection tools (Schmidt & Hunter, 1998; r ≈ .44–.54 for general work samples, higher for highly representative samples). They evaluate the work, not the person — directly testing the candidate's ability to produce the output the job requires. ## Design Principles 1. **Representative**: mirrors actual work, not abstract puzzles 2. **Bounded**: 60–120 min onsite; take-homes capped (3 hours common; compensated above ~3 hours) 3. **Same prompt** for every candidate at the same level 4. **Multi-evaluator**: 2+ reviewers with shared rubric 5. **Discussed, not just produced**: candidate walks through their work and reasoning ## Role-Specific Patterns ### Software Engineer - Coding task on a representative codebase or a clean problem with realistic constraints - Debugging exercise on real-looking code - Design discussion of an actual recent system decision - Pair-programming on a real bug or feature - Avoid: abstract algorithmic puzzles uncorrelated with the work ### Designer - Critique of existing work (yours or theirs) - Small design task with explicit constraints - Portfolio walkthrough with deep questioning - Working session on a real problem ### Product Manager - Analysis of a real product decision the company made - Hypothetical case mirroring an actual challenge - Roadmap exercise with constraints - Customer interview synthesis exercise ### Sales / Account Executive - Pitch a product (theirs or yours) to a panel - Discovery call role-play with realistic objections - Account plan for a hypothetical customer - Forecast discipline conversation ### Customer Success / Support - Customer call role-play with escalation - Account health analysis - Difficult customer email drafting exercise - Renewal conversation role-play ### Marketer - Channel strategy doc for a hypothetical - Writing exercise (blog, email, ad copy) - Analysis of a recent campaign with critique - Persona / ICP exercise ### Recruiter - Role intake call role-play with hiring manager - Sourcing exercise on LinkedIn / GitHub - Candidate brief writing - Difficult candidate conversation ### Executive / Senior Leader - 30-60-90 plan based on a brief - Org design review of a real or hypothetical structure - Customer or board presentation - Difficult conversation role-play - References-based deep-dive ## Scoring Pre-defined rubric per work sample with: - 3–5 evaluation dimensions (e.g., quality, judgment, communication, technical skill) - Anchored ratings per dimension (what does 4/4 look like?) - Multiple evaluators independently score before debrief - Specific evidence cited ## Fairness Considerations - **Time bounds matter for equity**: long take-homes disadvantage candidates with caregiving responsibilities, second jobs, less flexibility - **Compensate substantial take-homes**: signals respect; reduces self-selection bias - **Don't assume universal context**: don't ask "redesign Stripe's checkout" if the candidate hasn't used Stripe - **Accommodations**: candidates with disabilities may need additional time or different format - **Distinguish skill from style**: a less polished output that demonstrates strong reasoning may be better than a polished one with shallow thinking ## Common Failures - Work samples that don't represent the job (algorithmic puzzles for product engineering) - Take-homes that grow to 10+ hours - No standardization across candidates - Single evaluator scoring - Work samples evaluated by people who can't actually judge the work - Using the work product (the candidate gives away IP) ## Take-Home vs. Onsite | Option | Pros | Cons | |--------|------|------| | **Onsite** (60–120 min) | Time-bounded; consistent conditions; less candidate burden | Pressure may not reflect real work conditions | | **Take-home** (capped) | Reflects realistic work conditions; allows depth | Time disparity across candidates; risk of unpaid labor | | **Pair / live coding** | Reveals reasoning and collaboration | Performance under observation may not generalize | Most loops benefit from a combination. ## Key References - Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology. - Bock, L. (2015). *Work Rules!* - Industry practice from technical hiring guides (e.g., interview frameworks at Stripe, GitLab, Atlassian)