A good phone screen is a short, structured first interview that answers one question: does this candidate clear the bar for a serious conversation? Run well, it takes 20 minutes, asks every candidate the same three or four rubric-mapped questions, and produces a scored, justified record instead of a gut feeling. Run badly, it's an unstructured chat that costs an hour of calendar time and returns a vibe.
This guide is the how-to. It gives you a minute-by-minute phone interview script, a question bank grouped by what each question tests, and a four-criterion scoring rubric you can copy today. It's the fix-side companion to the seven ways phone screens fail; everything below exists to keep your screen out of that article.
What does a good phone interview script look like, minute by minute?
As of July 2026, practitioner convention puts a screen at 15–30 minutes; this structure uses 20, which is enough to cover four criteria without drifting into a full first-round interview. The script below is the citable version: copy it, then keep it identical for every candidate in the role.
The 20-minute phone screen structure
| Minutes | Segment | What happens |
|---|---|---|
| 0–2 | Set the frame | Introduce yourself, state the agenda and duration, tell the candidate you'll be taking notes. Two minutes of warmth, not ten. |
| 2–5 | Practical fit | The factual checks: notice period, compensation expectations, location and working hours, work authorization. Mismatches surface now, not in round three. |
| 5–16 | Core questions | Three or four planned, rubric-mapped questions with follow-ups. This is the interview. Protect it. |
| 16–19 | Candidate questions | Their turn, with real answers, including the honest ones. Candidates are screening you back. |
| 19–20 | Close | State the next step and when they'll hear from you. Then actually do that. |
Two design choices deserve a defense. The rapport segment is deliberately short because first impressions are not harmless: in a study of 163 structured mock interviews (Swider, Barrick and Harris, Journal of Applied Psychology, 2016), impressions interviewers formed during pre-question rapport correlated with their final scores and partly measured things unrelated to the job. And logistics come early because a fatal mismatch on salary or start date makes the remaining fifteen minutes a courtesy; better to know at minute five.
Put the format in the invitation: duration, that questions cover their real work, and that notes are taken. Candidates who know the shape of the call prepare evidence instead of small talk, which raises the signal of the eleven minutes that matter.
Immediately after the call, spend two minutes scoring while the evidence is fresh. That's the whole script: about 25 minutes of real cost per candidate, before scheduling overhead.
What questions should you ask in a phone screen?
Ask few, ask the same ones, and ask about the past. The research behind those three rules is consistent: structured interviews with fixed questions and anchored scoring are among the strongest predictors of job performance (operational validity .42 versus .19 for unstructured conversations; Sackett, Zhang, Berry and Lievens, Journal of Applied Psychology, 2022), and meta-analytic comparisons find past-behavior questions ("tell me about a time you...") add predictive power beyond hypotheticals ("what would you do if..."; Taylor and Small, Journal of Occupational and Organizational Psychology, 2002). Structured follow-up probing on those answers is part of what the structure literature endorses (Levashina, Hartwell, Morgeson and Campion, Personnel Psychology, 2014). The question-design rules go deeper; here they're applied.
The phone screen question bank. Ten questions, grouped by what they test. Pick three or four per role, map each to a rubric criterion, and keep the wording fixed across candidates. Each comes with what a strong answer includes, which doubles as your scoring anchor.
Questions for any role
1. "Walk me through the piece of work you're proudest of from the last year. What was your specific contribution?" A strong answer names a concrete artifact, separates their role from the team's, and includes at least one tradeoff they personally made.
2. "Tell me about a time you disagreed with how something was being done. What did you do?" A strong answer has a real disagreement, a direct action (not silent compliance), and an outcome where the working relationship survived.
3. "What does your current role look like day to day, and which parts do you want more or less of?" A strong answer is an honest split that maps onto what this role actually offers, not a flattering mirror of the job ad.
Questions for engineering roles
4. "Describe the last system you built or significantly changed. What broke, or nearly broke, and why?" A strong answer explains the architecture in plain language, names a real failure mode, and shows what they'd do differently now.
5. "Tell me about a technical decision you made that you'd reverse today." A strong answer owns a genuine tradeoff, explains what changed, and blames context rather than colleagues.
Questions for sales roles
6. "Walk me through a deal you won that you almost lost. What turned it?" A strong answer has deal mechanics: the objection, the specific action that changed the outcome, and numbers that add up.
7. "Which target did you miss most recently, and what did you change afterward?" A strong answer owns the miss with figures, names one concrete change, and shows evidence the change stuck.
Questions for ops and support roles
8. "Describe a process you inherited that was broken. What did you change first, and why that first?" A strong answer shows diagnosis before action, a defensible sequence, and a measured result.
9. "Tell me about the hardest customer or stakeholder situation you've resolved. What made it hard?" A strong answer walks through de-escalation steps and holds a clear line between empathy and policy.
The motivation question (any role)
10. "What would need to be true about your next role for you to stay three years?" A strong answer is specific enough to check against this role, including the parts that don't fit. Flattery is a weak answer.
How do you score a phone screen?
With a rubric small enough to use at speed. Four criteria cover what a 20-minute call can actually evidence; the full method for writing weighted criteria is in how to write a hiring rubric, and this is that method cut down for screens.
A four-criterion phone screen rubric
| Criterion | Tier | What 8–10 sounds like | What 0–3 sounds like |
|---|---|---|---|
| Core capability signal | Must Have | Concrete artifacts, real tradeoffs, their own role clearly separated from the team's | Generalities and buzzwords; no example survives a follow-up |
| Motivation for this role | Very Important | Reasons specific to this role, stage, and company | Any-job answers; motivation is purely escape from the current role |
| Communication | Important | Structured answers in plain language; answers the question that was asked | Wanders, hides behind jargon, or answers a different question |
| Practical fit | Important | Compensation, notice, location, and hours all workable | A hard mismatch on a non-negotiable |
Score each criterion 0–10 with a one-sentence justification citing something the candidate said. Score in the two minutes after the call, not at the end of a five-call day. And never score what you didn't observe: if a criterion never came up, mark it unscored and exclude it from the average instead of defaulting to a middle number. Weighted by tier, four scores and four sentences per candidate are enough to rank a screen pool defensibly, and the justifications are what make Monday's debrief an argument about evidence instead of memory.
Calibrate after the first ten screens: a criterion where everyone scores 7 or better isn't discriminating and needs a harder evidence anchor, and a criterion you keep leaving unscored means its question isn't earning its minutes.
Who should run the phone screen?
Whoever runs all of them for the role. Consistency beats seniority here: a recruiter who runs every screen with the same script produces more comparable results than a rotating cast of hiring managers improvising, because the value of a screen is relative ranking, and ranking needs a fixed instrument. The hiring manager should design the questions and evidence anchors with the recruiter, then trust the scores. If the hiring manager insists on running screens personally, cap the load: quality decays across a day of back-to-back calls, and the fifth candidate deserves the same interviewer the first one got. Three or four screens a day with scoring breaks between them is a sustainable ceiling.
What are the most common phone screen mistakes?
The CV recital. "Walk me through your CV" spends your eleven core minutes on information you already have. Ask about the CV's most load-bearing claim instead.
Winging it. Improvised questions turn your screen into an unstructured interview, and unstructured is measurably the weaker instrument (.19 versus .42, per the Sackett re-analysis above). The script is the fix; keeping to it is the discipline.
The 70/30 inversion. If you talk for most of the call, you screened yourself. Pitch briefly at the close; the middle belongs to the candidate's evidence.
Scoring from memory. Evening scoring after six calls is where consistency dies. Two minutes per call, immediately after, with justifications.
The rapport overrun. Ten minutes of warm-up feels kind and quietly biases the next ten (that's the Swider finding above). Two minutes, then structure. Warmth belongs in tone, not in duration.
Skipping logistics. Screens that dodge compensation and notice produce finalists who evaporate in offer week. Minute two to five exists for a reason.
Can you run the same screen without the phone call?
Yes, and the research quietly predicted it. A meta-analysis of a structured interview conducted by telephone and scored later from the taped transcript found essentially the same validity as conventional structured interviews (Schmidt and Rader, Personnel Psychology, 1999). The predictive power lives in the fixed questions, the evidence anchors, and the disciplined scoring, not in the live phone line or the interviewer's presence.
That is exactly the substitution async screening makes. (Disclosure: Rubrily is our product, so this is the part where we have a horse in the race; everything above works on paper without us.) In Rubrily, the same four-criterion rubric drives an async AI interview: every applicant gets the same questions with adaptive follow-ups, on their own schedule, in their own language; every answer is transcribed and scored 0–10 per criterion with a written justification; and the interview blends with the CV evaluation into one explainable Fit Score. The difference from the manual version isn't the method, it's the coverage: you screen every applicant this way, not just the fifteen your calendar could hold, and the record arrives already written.
FAQ
What questions should I ask in a phone screen? Three or four fixed, past-tense questions mapped to your rubric: one core-capability question, one collaboration or judgment question, one role-specific craft question, and one motivation question, plus factual logistics checks. Keep wording identical across candidates so scores stay comparable.
How do you score a phone screen? Against a small rubric: four criteria (capability signal, motivation, communication, practical fit), each 0–10 with a one-sentence justification, scored immediately after the call. Leave unobserved criteria unscored rather than guessing a middle number, and rank candidates by the weighted result.
How long should a phone screen be? Convention is 15–30 minutes; 20 is enough for four criteria: two minutes of framing, three of logistics, eleven of planned questions with follow-ups, three for the candidate's questions, one to close. All-in cost with scoring is about 25 minutes per candidate.
Who should run the phone screen? One consistent person per role, usually the recruiter, using questions the hiring manager helped design. Consistency of interviewer and script is what makes screen scores comparable; cap anyone's load at three or four screens a day to keep the last call as fair as the first.
Or skip the calendar entirely: the same rubric, every applicant interviewed async in their own language, every answer scored with a written reason. Get the free rubric template, or join the waitlist →