Removing a candidate’s name from a work sample can change what reviewers see. It does not, by itself, define what good work looks like. Without a shared scoring guide, reviewers may simply replace one set of impressions with another.
Name the skill being assessed
Before collecting samples, describe the task and the qualities that matter. For a short customer reply, those might include factual accuracy, clarity and an appropriate next step. Avoid vague criteria such as “sounds like us” unless you can explain the job-related behaviour behind them.
Create a few sample responses and ask reviewers to score them independently. Discuss disagreements before real applications arrive. If one reviewer rewards elaborate wording while another rewards brevity, clarify what the task actually requires.
Keep the process comparable
Give candidates the same essential information about the task, expected time and assessment criteria. Provide a clear route to request an adjustment. A timed exercise may unintentionally test speed with a particular tool when the role mainly requires thoughtful analysis; choose the format deliberately.
Keep identity information separate where the process intends anonymity, but be honest about its limits. Writing style, examples or filenames may reveal clues. Do not promise that anonymisation removes all bias or makes an assessment automatically fair.
- Record a score against each stated criterion.
- Add a brief evidence note from the work itself.
- Separate uncertainty from a negative judgement.
- Review unusual score differences before making the decision.
After the exercise, examine whether the task produced useful evidence for the actual role. If reviewers needed information the instructions never requested, improve the next version instead of penalising candidates for failing to guess it.
The strongest assessment record explains why the work met or missed a requirement. An anonymous filename can support that process, but it cannot substitute for a requirement that reviewers understand consistently.