How to interview a candidate

The seven steps

1. Write down the three things this role genuinely requires.
2. Write three or four questions that test those things, and nothing else.
3. Agree in advance what a weak, adequate and strong answer sounds like.
4. Ask the same questions to every candidate for the role.
5. Follow up until you reach a specific decision the candidate made.
6. Write down what they said, not what you concluded.
7. Score against the rubric before the debrief, independently.

Before the interview: decide what you are measuring

The most common failure in interviewing happens before anyone speaks. If you have not decided what the role requires, you will assess what the candidate happens to talk about, and the candidate who is best at steering the conversation will win. Write down three requirements — not ten, and not “strong communication skills”, which is a quality rather than a requirement. Something like: can own a service end to end, can debug a problem they have not seen before, can work with a product manager without an intermediary. Each should be specific enough that you could imagine evidence for it.

Structure the hour

• 5 minutes — Introduce yourself, the role and the format. Tell them how long you will spend on questions and when they will get time for theirs. Candidates perform better when they know the shape of the conversation, and you want their best, not their most anxious.
• 35 minutes — Three or four questions with follow-ups. Expect one question to take fifteen minutes if it is going well; let it.
• 10 minutes — Their questions. This is assessment too: what someone asks reveals what they care about.
• 5 minutes — Next steps, with a date. Never end on “we will be in touch.”

In the room: the 20% rule

You should be talking for roughly a fifth of the interview. Interviewers systematically talk too much — explaining context, filling silences, validating answers — and every minute you spend is a minute of evidence you did not collect. Silence after a question is productive; give it four or five seconds before rescuing the candidate. Most people fill it with the more honest half of their answer.

Follow up until you reach a decision

The first answer to any question is a summary. The evidence is one or two follow-ups deeper. A reliable sequence is: *what did you specifically do?*, then *what else did you consider?*, then *what happened afterwards?* Keep going until you reach a point where the candidate chose between two options and can tell you why. If you never reach that point, you have heard a narrative, not a decision, and narratives are not comparable between candidates.

Take notes that are evidence, not conclusions

• Write what they said, in their words where you can. “Chose Postgres over DynamoDB because the access patterns were not known yet” is evidence.
• Not what you concluded. “Good system design instincts” is an opinion, and an opinion recorded in the moment is indistinguishable in the debrief from an opinion formed afterwards.
• Note the questions you asked, especially follow-ups. If two interviewers reach different conclusions, the difference is usually in what they asked, not in the candidate.
• Note what you did not get to. Gaps are useful information for the next interviewer in the loop.

The mistakes that quietly ruin interviews

• Deciding in the first five minutes and spending the next fifty-five confirming it. This is the best-documented failure in interviewing and it feels exactly like good judgement from the inside.
• Interviewing without preparation because you know the domain well. Domain knowledge is not a rubric.
• Asking different questions to different candidates and then comparing scores as if they meant the same thing.
• Discussing the candidate before everyone has scored. The first opinion voiced anchors the rest, which destroys the independence that made a panel worth having.
• Treating nerves as data. Interviews are artificial and stressful. The question is whether they can do the job, not whether they enjoy being assessed.

Score before you discuss

Write your assessment within ten minutes of the interview ending, against the rubric you agreed beforehand, and before you talk to any other interviewer. This single habit does more for hiring quality than any question you could add. See how to write an interview scorecard and how to run an interview debrief.

Where Interviewer Mode fits

Disclosure: we make CoPilot Interview, which includes Interviewer Mode for the hiring side and a real-time assistant used by candidates. We have an interest in you using the first, and our view of the second is that interviews should be designed to measure something an assistant cannot fake. Weigh the advice below accordingly.

Frequently asked questions

How do I interview a candidate properly?

Decide what the role requires before the interview, write three or four questions that test exactly those things, agree what a strong answer sounds like, and ask every candidate the same set. In the room, talk about a fifth of the time, follow up until you reach a specific decision the candidate made, record what they said rather than what you concluded, and score independently before any debrief.

How long should I spend talking in an interview?

About 20% of the time. Interviewers routinely talk far more than that, and every minute you spend explaining context is a minute of evidence you did not collect. Leave four or five seconds of silence after a question — most candidates fill it with the more honest half of their answer.

What should I write down during an interview?

What the candidate actually said, in their words where possible, plus the questions you asked. Avoid recording conclusions like “strong communicator” — in a debrief, an opinion written during the interview is indistinguishable from one formed afterwards, and it crowds out the evidence.

How soon should I decide after an interview?

Score within about ten minutes of the interview ending, while the detail is still available, and before speaking to any other interviewer. Deciding later means deciding on impressions; deciding after a conversation means deciding on someone else's.

What is the most common interviewing mistake?

Deciding in the first few minutes and spending the rest of the interview looking for confirmation. It is well documented, and from the inside it feels like good instinct rather than bias. Structure and a pre-agreed rubric are the only reliable defences.