Structured, fair screening

Structured screening means every candidate for a role is asked the same questions and judged against the same standard, written down before anyone is interviewed. It's fairer to candidates, it keeps the conversation on the job, and it makes your decisions easy to explain.

Ask everyone the same questions

• Agree the questions for each stage before the first interview, and ask them in the same order.
• Keep every question about the job: what the person has done, how they'd approach the work, and what they're looking for.
• An AI screen does this by design. The interviewer asks your questions in order, to every candidate, in the language they choose.
• For live interviews, use the same note template for every candidate at a stage, so the notes line up when you compare them. Behavioural interview and Technical interview are good starting points.

Write the rubric first

For every question, decide what a strong and a weak answer look like before you see any answers. Once you've met a candidate you like, it's too easy to define a strong answer as whatever they said.

In an AI screen, each question has:

• Reveals: what a good answer tells you. If you can't fill this in, the question probably doesn't belong.
• Strong and Weak: what each kind of answer includes or looks like.
• A number of minutes, and whether to ask follow-ups.

A screen the agent drafts has four questions and takes about 13 to 15 minutes. Emoti warns that over 25 minutes is a lot to ask before a first conversation. See Setting up an AI voice screen.

Follow up for evidence, not to steer

Follow-ups help when an answer is thin: "What did you do yourself?", "What happened next?", "How did you know it worked?". Use them the same way with everyone.

• In an AI screen, turn follow-ups on for a question and the interviewer follows up when an answer is thin.
• Avoid leading follow-ups, such as "So you'd say you're a strong communicator?"
• Don't skip follow-ups for candidates who seem promising, or pile them on for ones who don't.

Score independently, then compare

• Each interviewer rates every answer against the rubric, and writes down the evidence, before talking to anyone else about the candidate. Hearing someone else's view first tends to pull your rating towards theirs.
• In an AI screen, each answer is rated Weak, Mixed, Good or Strong against your rubric, with a short summary, and the scorecard has an overall score from 0 to 100. People review the answers and decide; Emoti doesn't advance or reject anyone.
• Read the evidence, not only the score. A Mixed rating with a strong example deserves a second look. See Reading screen results and scorecards.

Run a debrief that ends in a decision

1. 1.Start with each interviewer's vote and main reason, given before any discussion.
2. 2.Discuss evidence against the rubric, not impressions. Separate open concerns from real blockers.
3. 3.Agree the decision, who owns each next step, and by when, before everyone leaves.

Record the debrief and write it up with the Debrief template: votes, strengths, concerns, and the decision and next steps. A debrief with only your teammates isn't joined automatically, so turn note-taking on for it under Upcoming. Afterwards, Ask can answer "What's the decision?" or "What concerns came up?" with links to the moment.

To see whether interviewers are listening more than they talk, check interviewer patterns in Insights across conversations.