A hiring exercise should test judgement without borrowing a client's secrets

A request for an example of previous work can sound entirely reasonable in an engineering interview. The hiring team wants to see how a candidate reasons, communicates and handles a complicated assignment. A polished finished study appears to answer those questions more convincingly than a conversation about experience. Yet the request may also ask the candidate to cross a professional boundary before joining a firm that will later expect them to defend the same boundary for its own clients. I think HR should notice that contradiction before it becomes an informal test of enthusiasm.

At Velqemi we had invited candidates to bring earlier client work into the recruitment discussion. The intention was practical assessment, not a deliberate search for confidential material. That intention did not settle what candidates were permitted to provide. A person with access to an impressive document could appear stronger than someone who refused to disclose it. The refusal might reflect better professional judgement, but the process could make it look like a lack of evidence. We needed to change the conditions of the assessment rather than ask interviewers to compensate informally for an unsuitable request.

The convenient request carries a hidden condition

A finished client deliverable contains more than the candidate's individual skill. Other people may have contributed, the client may own relevant material, and the conditions governing disclosure may be specific to the engagement. Removing a name or a logo does not necessarily resolve those matters. HR should not improvise a permission conclusion from the appearance of a document. Nor should a candidate be expected to decide under interview pressure that material is safe to share because a prospective employer has asked for it. The safer question is what evidence the firm can create or request on a proper basis.

There is also a problem of comparability. Two previous assignments may differ substantially in scale, available support, complexity and the candidate's own contribution. A highly polished presentation can conceal a limited role, while a less attractive example may reflect substantial independent judgement. Interviewers can explore those differences, but the document alone does not remove them. A recruitment process should explain which capability it is trying to observe and why the requested evidence is suited to that purpose. The impressive appearance of a prior engagement is not a substitute for that reasoning.

The hiring manager's objection to a fictional exercise deserved a serious response. He worried that candidates could rehearse generic answers and that the assessment would become detached from the work. I did not dismiss that concern as resistance to a people policy. Technical specialists needed to design a problem with meaningful choices, incomplete evidence and an opportunity to explain assumptions. HR could help with administration and consistency, but engineering leaders had to judge whether the exercise revealed relevant professional reasoning. Protecting confidentiality should not become an excuse for a weak assessment that leaves the actual appointment risk unexamined.

Build evidence the firm is entitled to assess

We developed a common fictional case and made the criteria explicit. Candidates could discuss their past responsibilities without bringing a previous client's files. The new exercise cost ₹27 lakh to develop and administer. That cost mattered because it represented work the firm had previously expected candidates to supply through their existing experience. A proper assessment required us to create relevant conditions ourselves. It was reasonable to ask whether the investment was proportionate, but it was not reasonable to call the older process costless simply because the preparation had occurred somewhere else.

Forty-six candidates entered the process. Thirty-eight completed the exercise and eight withdrew. Of the completed submissions, twenty met the stated criteria and eighteen did not. Those categories were kept separate. We did not call the eight withdrawals failures, because the absence of a completed submission did not tell us why somebody chose not to continue. We also did not describe the twenty satisfactory assessments as hires. An assessment result informs an appointment decision; it does not replace the other conversations and checks that the company needs before making an offer.

A common exercise needs a reasonable scope. An employer should not turn recruitment into an unbounded request for free consulting or expect candidates to solve a live client problem without an appropriate arrangement. The fictional nature of the case helped preserve that distinction. We also needed a way for candidates to raise practical concerns about the format. A technically useful problem can still be administered poorly. Clear instructions, an understandable time commitment and consistent criteria support the assessment, while leaving specialist judgement about the engineering response with the people qualified to make it.

The exercise did not reproduce the full experience of an actual engagement. Candidates were not working with the same colleagues, pressures or access to client knowledge they would encounter after joining. I wanted that limitation acknowledged rather than hidden behind a claim of scientific precision. The purpose was to obtain a useful piece of evidence under defensible conditions. Experience discussions, references and subsequent employment decisions remained separate. Combining several appropriate sources of evidence is more credible than asking one exercise to prove every dimension of a person's future contribution.

Keep the result smaller than the hiring promise

One risk after redesigning a process is that its owners become invested in declaring it successful. We had spent money and specialist time, so there was pressure to show that the new method produced better employees. The initial numbers could not establish that conclusion. They described participation and results against the defined criteria. A later account of employee performance would require its own evidence and a careful explanation of what could reasonably be attributed to recruitment. I did not want a principled change in assessment conditions to become an unsupported causal claim about the business.

The stronger immediate result was that candidates no longer had to choose between protecting previous work and satisfying our evidence request. Interviewers had a common case to discuss and a clearer account of what they were assessing. The firm paid for that improvement through development time, administration and the loss of convenient finished examples. Some candidates still withdrew, and some completed work did not meet the criteria. Those outcomes belonged in the account. A process can be more defensible without becoming frictionless or producing an ideal result for every participant.

For a board, I would frame the issue as consistency between the professional conduct a company promises and the conduct its employment practices reward. Directors do not need to design interview exercises. They can ask management whether hiring incentives undermine confidentiality, whether the evidence is suitable for the role and whether reported results distinguish assessments from appointments. My own lesson is narrower and practical: before asking a candidate to demonstrate judgement, make sure the request itself demonstrates the judgement the company hopes to employ.