Over the past six years, Metaview has captured behavioral data from tens of thousands of hours of real interviews.
Three patterns repeat for new interviewers. They have almost nothing to do with role, industry, or seniority of the hire. They show up at the same rates across an engineering loop and a sales screen.
Most teams try to coach them out. Few catch them in time.
This post names the three mistakes, shows the data, and gives the exact language to coach against each one.
1. Forgetting to set expectations
Most new interviewers cover a personal introduction. Almost none cover what the interview is going to do.
Across the Metaview corpus, only 28% of new interviewers set expectations at the start. The other 72% open with their own background, ask the first question, and the candidate spends the next ten minutes guessing what's being evaluated.
Setting expectations means the structure and purpose are clear in the candidate's mind from the opening minute.
It reduces anxiety around when the candidate will get to demonstrate their capabilities, so the early answers come out cleaner.
It reduces anxiety around how they're being judged, which improves the candidate's sense of closure at the end. It also reminds the interviewer what they need to cover, so they don't drift into a conversation that fills time but doesn't earn signal.
After a quick introduction, say: "The aim of this interview is to learn more about [x]. I'll [y], and I'd love for you to [z]. We'll have [n] minutes for your questions at the end. Sound good?" Two examples below.
| Don't say | Say | Go deeper |
|---|---|---|
| Cool, let's get started. Tell me about yourself. | The aim of this interview is to learn about your experience as an Account Executive. I'll ask about specific situations you've handled, and we'll save 10 minutes for your questions at the end. | I'll go deep on two or three real deals you've worked, including how you handled stakeholders and what you'd do differently. We'll have 10 minutes for your questions at the end. Sound good? |
| I'll ask a few technical questions, then a behavioral section, then we'll wrap. | The aim is to learn how you break down complex problems. I'll lay out a couple of scenarios, and I'd love for you to approach them as if we were teammates working it out. | I'll lay out two scenarios you'd see in this role. Treat me like a teammate, think out loud, ask whatever you need. We'll have 10 minutes for your questions at the end. Sound good? |
2. Being too polite to get concrete
Every interviewer is taught the goal is mutual fit. Few new interviewers push past the polite version of that conversation.
In the Metaview corpus, 83% of new interviewers don't ask for a single concrete example across the entire interview. Not one.
The whole conversation runs at the level of approach and philosophy. Past behavior is still the best predictor of future performance, and most loops never collect any.
The reason is anxiety, not laziness. Asking for a real example feels invasive. Picture-book answers feel safer to the new interviewer than they do to the candidate.
Vague questions get answered vaguely. Both sides shake hands at the end feeling fine about it, and nobody learns anything that would predict performance on the job.
If you're asking about something the candidate should have experience with, the question is about an experience, not an opinion. The harder it is to invent a clean answer, the closer you are to the signal you actually need.
| Don't ask | Ask | Go deeper |
|---|---|---|
| How do you manage stakeholders? | What techniques have worked well for you when managing lots of stakeholders? | Can you walk me through a specific time you had many stakeholders to manage, and how you went about it? |
| What sort of teams do you like being part of? | What sort of teams have you most enjoyed being part of in the past? | Think about a team where you were performing at your best. What were the key characteristics of that team and your role in it? |
3. Making too many assumptions
Closed and leading questions slip in because the interviewer is making assumptions, often based on their own career.
In the Metaview corpus, new interviewers ask 50% more closed or leading questions than experienced interviewers. Half again as many questions where the answer is already in the question.
That single behavior does more damage to a hiring loop than the other two combined, because it's invisible from the outside.
The candidate gives back the answer they think you want, which is rarely the actual answer.
The interviewer's preconceptions get confirmed rather than challenged, which calibrates the team in the wrong direction. Phrase the question so the candidate provides the framing, not just confirms yours.
| Don't ask | Ask | Go deeper |
|---|---|---|
| Did you make that decision based on some feedback you'd received? | How did you come to that decision? | Walk me through how you arrived at that decision, what trade-offs you weighed, and what you'd do differently now. |
| I guess you have a lot of experience working in regulated environments? | Can you tell me about your experiences working in regulated environments? | Talk me through a recent project in a regulated environment and how the constraints shaped what you decided to ship. |
Looking at all four behaviors together, across hundreds of customer teams and six years of data, the gap between new and experienced interviewers is structural, not accidental.
How Metaview captures this
Question count, question type, candidate speaking time, time spent on context, interruptions, whether expectations were set at the open. All of it is auto-extracted by Metaview's Notetaker from a recorded interview. No interviewer fills in a form.
Once a team has a few weeks of interviews captured, the patterns stop being industry statistics and become specific questions about your team. Which interviewers are in the 28% that set expectations? Which are in the 83% that never get concrete? Whose questions are leading?
AI Filters answers that in plain language. Type "show me last month's interviews with fewer than three behavioral questions" and the list returns, ranked, with the recordings one click away.
We elevated from gut-feel recommendations to evidence-based insights, creating a faster, clearer, and more data-driven experience for everyone involved. Every scorecard and report looks and sounds consistent, regardless of who prepared it."
The broader story matters. Across our 2026 AI & Hiring Alignment Report, surveying 505 recruiting leaders and hiring managers across North America and EMEA, 85% of companies exceeding their hiring goals use AI in hiring. The teams pulling ahead aren't using AI to do more interviews. They're using it to surface and coach the upstream behaviors that decide whether the interviews they already run are worth running.
How hiring improves when these changes are made
Coaching new interviewers used to be a side project. Someone senior would shadow a screen now and then, take notes, share feedback a week later.
By that point the interviewer had already done four more screens with the same mistakes in them.
When every interviewer's question-type distribution is sitting in Metaview Reports, the conversation between an interview lead and a new interviewer changes.
It's no longer about whether the gap exists. It's about which two clips to listen back to and what to try differently on the next call.
Calibration audits stop being quarterly and start being weekly.
If three of the five interviewers in your loop are still in the 83%, the screening problem you think you have isn't a screening problem. It's a coaching one, and it's solvable in a few weeks once the data is in front of the right person.
Frequently asked
What are the most common mistakes new interviewers make?
Three repeat across the Metaview interview corpus. Only 28% of new interviewers set expectations at the start of an interview. 83% don't ask for a single concrete example across the whole conversation. And they ask 50% more closed or leading questions than experienced interviewers. Each one weakens the signal the loop is trying to collect, and they show up at the same rates across role, industry, and seniority of the hire.
How do I become a better interviewer?
The fastest path is feedback on real interviews, not training decks. Record your calls, look at your own question patterns (how many were closed, how many were behavioral, whether you set expectations), and have someone senior listen back to two specific moments per call. Metaview's AI Filters surfaces those patterns automatically so the feedback loop closes in days, not quarters.
What's the best way to coach a new interviewer?
Pair shadowing with structured debriefs from real interview data. Watching one live call is useful. Pulling clips of the same interviewer across ten screens and showing them their question-type distribution is better. Metaview Notetaker captures every interview automatically, AI Filters surfaces the patterns, and you go from a generic "ask more behavioral questions" note to "in screen #3 with the candidate from Acme, here's the moment you should have probed."
How can I tell if my team's interviews are actually working?
Look at the upstream behavior, not just the funnel. If your offer-acceptance rate is dropping or candidates are ghosting you mid-loop, the three mistakes above are usually upstream of it. The same gaps that lose signal also lose candidates. Capture two weeks of interviews, run the questions above against your own data in Metaview, and you'll know in an hour where to start.
Bring Metaview into your hiring stack.
Live notes, structured scorecards, and ATS sync - set up in under 10 minutes.
