Philippines staffing guide
Design a Paid Work Sample for an Offshore Role
Use a real, sanitized task with a rubric, a time limit, and the same tools the role will use. Score evidence, not presentation.
The short answer
A work sample is useful when it is real, bounded, scored against a rubric, and paid if it produces value.
Choose a task from the real queue
Select a task the role will actually perform in the first month. Use sanitized data or a synthetic case that preserves the decisions. Avoid puzzles and trivia.
The sample should show whether the candidate can find the source, follow a rule, record evidence, and stop when something is missing. If the role involves customer writing, include a customer message. If it involves spreadsheets, include a reconciliation with one deliberate discrepancy.
Bound the time and the scope
Give a clear time limit and a clear finish line. A task that takes four hours is not a sample. It is unpaid work.
Decide whether the candidate may use the tools they would use on the job and whether they may ask questions. Tell them how to flag uncertainty.
A candidate who spends the whole limit polishing the formatting and misses the discrepancy has shown something useful about prioritization. That is the point.
Score against a written rubric
Use four or five dimensions: accuracy, source use, evidence quality, communication, and boundary awareness. Define what a strong, acceptable, and weak response looks like for each. Score independently before discussing the candidate with the other interviewers.
Keep the completed sample and the score sheet for the onboarding plan. If the candidate is hired, the sample becomes a baseline for coaching. If not, the record supports a fair comparison.
Pay for substantive work
If the sample produces value the company could use, pay for it. A short exercise that is clearly for assessment may be unpaid in some contexts, but the boundary should be explicit and fair. Do not build a production backlog out of candidate submissions.
Do not ask candidates to solve a live customer problem without a confidentiality and payment arrangement. The hiring process is also a signal about how the company treats people.
Calibrate before the real round
Run the sample with a current team member or a colleague who is not a candidate. Check the time limit, the instructions, and the rubric. Remove ambiguity before the first candidate sees it.
After the first two candidates, review whether the rubric is discriminating on the right things. Do not change the rubric between candidates unless the earlier scores are re-reviewed. A consistent process is more valuable than a perfect one.
Questions to copy for the sales call
- "Which real task will the sample simulate?"
- "What is the time limit and the finish line?"
- "How will accuracy, evidence, and boundaries be scored?"
- "Will the sample be paid, and who owns the output?"
Sources
- International Labour Organization, Data: International labor standards and data context. Checked October 22, 2026.
- U.S. GAO, Assessing Data Reliability: Provenance and fitness-for-use questions. Checked October 22, 2026.
- World Bank, Philippines Overview: Country context for Philippines-based service work. Checked October 22, 2026.
Common questions
How long should a work sample take?
Usually under two hours for a screening sample. Keep it focused on the decisions the role will make, not on volume.
Can we use a real customer case?
Only with sanitization, a confidentiality arrangement, and a clear payment or assessment boundary. Never expose live customer data to a candidate.
Should candidates use AI tools?
Decide in advance and state it in the instructions. If the role uses approved tools, allow them and score the verification and evidence, not just the output.