An unstructured interview mostly measures how comfortable someone is in an unstructured interview. That is a real skill, and a job requirement for almost nobody.
A structured interview asks every candidate for a role the same core questions, in the same order, scored against criteria written before anyone was interviewed. It exists so two interviewers can compare answers rather than impressions — which is the only thing that makes a second opinion worth having.
Small teams resist this because it feels corporate, and because the founders can usually tell within ten minutes. The uncomfortable part is that they can — they are just as often wrong, and there is no way to find out which time this was.
An unstructured interview mostly measures how comfortable someone is in an unstructured interview. That is a genuine skill, and it is a job requirement for almost nobody. Structure does not make the conversation less human; it makes the comparison possible afterwards.
Step three is the one people skip, and it is the one that makes scoring mean anything. "Good communicator, 4/5" is not evidence. "Described the disagreement from the other person's side without being asked to" is.
If you would rather not start from a blank page, Generate a question set for the role
The core set is fixed. That is what makes candidates comparable, and it is also what protects you if a decision is ever challenged, because the alternative is explaining why one candidate was asked about gaps in their history and another was not.
Diverging is fine and necessary in two places: following up on what someone actually said, and probing something specific to that person's background. The rule is that the core is the same for everyone and the follow-ups are wherever the answers lead.
Behavioural questions beat hypothetical ones for the same reason. "How would you handle a difficult stakeholder" measures fluency at answering that question. "Tell me about the last time you did" produces a specific occasion you can ask three more questions about.
Immediately is doing the work in that sentence. An hour later you are recalling a feeling rather than an answer, and the reconstruction is reliably more generous to candidates you liked and less generous to the ones who were nervous early.
Record three things separately: what they were asked, what they said, and what you concluded. Keeping the conclusion apart from the evidence is what lets someone else disagree with your reasoning rather than just with your verdict.
Then, before the debrief, everyone submits their notes without seeing anyone else's. This costs nothing and removes the single most common failure of a hiring debrief, which is that the first person to speak sets the position everyone else adjusts toward.
It does not fix a role nobody has defined. If the requirements are vague, the scorecard is vague, and you have added a form to the same argument you were already having.
It does not make anyone objective, and a structured process run by people who all think alike produces consistent, well-documented, uniform judgement.
And it does not replace the decision. At the end there is a table of evidence and a person who has to decide, which is as it should be — the point of the structure was never to remove the judgement, only to make sure it was applied to the same thing for everyone.
Four to six substantive ones in an hour, with room to follow up. A list of twenty gets abandoned twenty minutes in, and what replaces it is the improvised interview you were trying not to have.
Yes, with two questions instead of six. The part that matters is that it is the same two for everyone and that someone writes down the answers.
Define what you mean before you assess it, or it becomes a licence to hire people like the people already there. If you mean specific working norms — how disagreement happens, how decisions get made — say so and ask about those directly.
For some roles it improves the signal considerably: you get considered answers rather than fast ones, and most jobs reward the former. It is worth trying for senior roles where preparation is part of the work.
Score the evidence against each requirement separately and never total it. A single number is where the information goes missing, because it stops you seeing that a strong overall score came entirely from things the role does not need.
Operated by Alano Tech Pte. Ltd., Singapore.
Terms of Service Privacy Policy
Also from us: Alano.ai Opptymizer Omu Labs