How to Run Structured Interviews
Run an interview that produces comparable, scored evidence for one role. Reach for it once the scorecard exists and before you meet the first candidate, so answers differ because of the candidates, not because they were asked different things.
Developing
Start here. Build the foundation.- 1
Pull two or three questions per assigned skill from the shared bank, or write them, and fix the order so every candidate for the role gets the same questions in the same sequence. Write each as a past-behavior question, 'Tell me about the last time you...' not 'How would you handle...,' so only someone who has done it can answer.
- 2
When a question targets a role skill, require an actual situation: when it happened, what the candidate personally did, and how it came out. If the answer is hypothetical ('I usually make sure to...'), ask again for one specific time and do not score the general version.
Proficient
Build consistency and rhythm.- 3
Right after the interview, before you talk to any other interviewer, score each skill against its anchor with a one-line note of the evidence: 'Strong: rebuilt the close process solo, cut it two days' beats a bare number. Record it on its own, because waiting until the debrief lets the room's mood rewrite what you heard.
- 4
After an answer, ask a neutral follow-up for the missing specifics ('What did you do next?', 'What was your part?') before you react, flat and without hinting which answer you want. A nod at the right answer teaches the candidate what to say next, and a thinking pause is working time, not a gap to fill.
Mastered
Operate at the highest level.- 5
After a search, add the questions that separated strong from weak candidates to the shared bank, tagged by skill, and retire any question every prepared candidate now answers the same way. A question everyone aces tests preparation, not the skill, so the next interviewer draws from a tested set rather than starting cold.
Common Pitfalls
Avoid the common failure modes.- Candidates got different questions because the set was not fixed. Lock the question order and reuse it for the whole role.
- Every answer is hypothetical because the questions asked 'how would you,' not 'when did you.' Rewrite them to demand a past instance.
- Everyone scored after the debrief, so impressions merged into one. Score and submit before anyone compares notes.
- A question the whole market has rehearsed no longer separates candidates. Retire it and replace it with one that still does.