Work samples: the assessment with the best predictive record
Want to know how someone will do the job? Watch them do a small piece of it.
Of all selection methods, work samples have among the strongest evidence for predicting actual performance - above interviews, references and certainly above gut feel. Yet most BFSI hiring still runs on conversation alone. The fix is cheap: give finalists a small, real piece of the work.
Designing a fair work sample
- Real but bounded: a credit memo on a sanitized case, a market-entry one-pager, a portfolio review of a masked book
- Ninety minutes to a half-day of effort - anything more is unpaid consulting and filters out the employed
- Same brief, same time, same materials for every candidate
Score the thinking, not the polish
The deck design is not the job. Score the structure of the argument, the quality of the questions the candidate asked back, the assumptions they made explicit, and whether their recommendation follows from their analysis. A plain memo with a spine beats a beautiful one without.
The candidate-experience rule
Never use work samples to harvest free work for live problems, and say so explicitly: "this case is historical and anonymized." Senior candidates are evaluating you too, and the ones you most want are precisely those who will decline to donate billable thinking to your current project.