The Recruiter’s Playbook: How to Design a High-Quality Voice AI Prescreen

Voice AI lets you talk to every applicant. But the value is only as good as your rubric + script + data mapping. If those are crisp, you’ll get ranked shortlists that hiring managers trust — and you’ll never miss qualified talent again.

The 12-Step Setup

1. Define the outcome for the role
   What does success look like in 90 days? Turn those outcomes into measurable signals.
2. Select the signals you’ll measure
   • Eligibility logistics: location, shift, authorization
   • Must-haves: certification/years/tools
   • Core skills: scenario questions
   • Intent/alignment: compensation range, timeline
   • Communication: clarity, structure, follow-through
3. Create knockout rules (objective only)
   Examples: “RN license active in state”, “Within 30 mi of warehouse”, “Can lift 50 lb”, “Authorization = Yes”. Knockouts should be binary and auditable.
4. Write 5–7 concise questions
   • 2–3 eligibility/logistics (30–45s)
   • 2–3 core skill prompts (90–120s total)
   • 1 scenario (60–90s)
     Keep it under ~7 minutes. Always tell the candidate what happens next.
5. Build the scoring rubric (weights)
   Example default for volume roles:
   • Role essentials (must-haves) 40%
   • Availability & logistics 25%
   • Communication & problem solving 20%
   • Intent/alignment 15%
6. Set thresholds
   • Pass: ≥ 75 → Shortlist
   • Review: 60–74 → Human review
   • Fail: < 60 or any KO → Reject with reason
7. Choose voice & language strategy
   • Primary: EN-US (Brand Female)
   • Fallbacks: ES-MX, FR-FR, HI-IN, etc.
   • Auto-detect or route by candidate locale. Keep tone friendly, concise, and brand-safe.
8. Consent & compliance
   • Open with consent line; log opt-outs.
   • Recording/transcript retention policy.
   • No sensitive/EEO-restricted questions.
   • Offer alternative channel if needed.
9. Outreach windows & retries
   • 2–3 attempts across time windows (e.g., 10–12, 17–19 local)
   • Optional voicemail + SMS/email follow-up with self-schedule link
10. ATS field mapping

• Statuses: “Phone Screen – Passed/Review/Rejected”, “No Response”
• Score → numeric field; Disposition reason; Tags (e.g., SW-shortlist)
• Attach summary PDF + transcript link (if policy permits)

11. Pilot & A/B test

• Run two versions of a question or voice and compare pass-through and no-response rates.

12. Monitor & iterate (weekly)

• Time-to-screen, KO rate, pass-through, shortlist quality delta, no-response %.
• Tune the question order, retry windows, or weights as needed.

Three ready-to-use templates

Use these as a starting point; edit the language to match your company voice.

1) Warehouse Associate (Night Shift)

Knockouts: Work auth, commute ≤30 mi, lift 50 lb, night shift availability
Questions:

1. “Are you available for Sun–Thu, 10pm–6am?”
2. “Tell me about your last warehouse role and the equipment you used.”
3. “How do you stay accurate when picking in a fast environment?”
4. Scenario: “You’re 10 orders behind at 4am. What steps do you take to catch up safely?”
5. “Earliest start date and pay expectations?”

Thresholds: Pass ≥75, Review 60–74
Tags: NightShift, Forklift, Accuracy-Strong, SW-shortlist

2) Sales Development Representative (Inbound/Outbound)

Knockouts: Work auth, hybrid 3 days/wk, CRM experience
Questions:

1. “Walk me through your prospecting workflow.”
2. “Share one example of turning a ‘not now’ into a meeting.”
3. Scenario: “2 emails + 1 VM → no reply. What’s your next move, and why?”
4. “How do you qualify inbound leads quickly?”
5. “What OTE range are you targeting and your earliest start?”

Weights: Core skills 35%, Comm 25%, Availability 15%, Intent 25%
Thresholds: Pass ≥78; Review 65–77

3) Registered Nurse (Clinic)

Knockouts: Active RN license (state), BLS, weekend rotation availability
Questions:

1. “Tell me about your recent clinical experience and patient load.”
2. “How do you handle triage during peak hours?”
3. Scenario: “Two high-priority patients arrive simultaneously — how do you decide?”
4. “Experience with EMR systems?”
5. “Shift flexibility and pay expectations?”

Compliance: Avoid medical history, protected categories; retain transcript per policy.

Openers, closers, & bilingual snippets

Intro + consent (EN):

“Hi, this is Screenwise calling about the [Role] at [Company]. With your permission, I’ll ask a few short questions to confirm fit. You can stop any time. May I proceed?”

Intro + consent (ES):

“Hola, le llama Screenwise sobre el puesto de [Role] en [Company]. Con su permiso, haré unas preguntas breves para confirmar el ajuste. Puede parar cuando quiera. ¿Puedo continuar?”

Closer:

“Thanks for your time — we’ll review your responses and send next steps by email or text. Have a great day.”

Dashboard: how to prove it’s working

• Time-to-screen (hours) — goal: <24h
• KO reason distribution — fix job ads if a single KO dominates
• Shortlist quality delta — Avg score (shortlist) vs (all applicants)
• No-response rate — tune windows, add SMS
• Pass-through — Prescreen → Interview → Offer

Show these in your main dashboard: Funnel, Time-to-Hire trend, Shortlist Quality bar, Candidate Feedback.

Common pitfalls (and easy fixes)

• Scripts too long → keep it under 7 minutes
• Vague questions → use scenarios with concrete, job-specific detail
• Threshold too high → adjust after 1–2 weeks of real data
• No-response spike → add evening/weekend windows + SMS reminders
• Manager distrust → surface key quotes + transcript snippets in ATS

The bottom line

A consistent, fair prescreen is the difference between noise and signal. With a simple rubric, concise questions, and clean ATS mapping, Voice AI helps your team hear from everyone, score them fairly, and move fast on the right people.

‍

‍

‍