The Business Case for AI Interviewing: What Every Screen Costs and Saves

TL;DR: You'll lose the CFO meeting with a deck. You'll win it with a model. This page builds the business case from the smallest unit up: what one completed screen costs today, what it costs automated, and how fast the delta pays back. Copy the boxes into a spreadsheet, plug in your numbers, and walk in with a payback period instead of a pitch.

You spend months evaluating vendors and one afternoon building the budget case. Then you lose in the finance meeting.

Your CFO has already sat through three AI pitches this quarter. Every one promised to "transform hiring." None handed over a model finance could check. The internal sell isn't won with features. It's won with math.

The business case for AI interviewing is a spreadsheet, not a story. Every section below is a boxed calculation you can copy directly, with Humanly-benchmark defaults so you're never stuck guessing. One scope rule: we only count costs and benefits that touch the interview. Recruiter minutes on screens, no-shows that waste booked slots, the days the screen-to-schedule gap adds to a vacancy. If a cost doesn't touch an interview, it doesn't go in this model. For the platform-wide case (sourcing through offer), see Humanly's AI Recruiter ROI 2026: The CFO-Ready Business Case. For what AI interviewing actually is, start with the first-round buyer's guide.

Executive takeaway: Vendors lose CFO meetings on story. You win them on a model finance can reproduce without you in the room.

Price the status quo before you price the software

Before you ask for budget, put a number on what you're already spending. Most teams lead with software cost, which frames the conversation as new spend. Reframe: you're not overspending on tools. You're paying for friction.

Here's the boxed calculation for your current interviewing line. Every term is interview-attributable.

Cost of the manual interviewing line (annual) Hires per year × screens per hire × minutes per screen × recruiter loaded cost per minute + interview scheduling & coordination drag + no-show losses (empty interview slots) + vacancy days added by the screen-to-schedule gap × daily vacancy cost = annual cost of manual interviewing

Here are benchmark defaults for each input:

• Screens per hire: 6–8 phone screens
• Minutes per screen: 20–30 min including notes
• Recruiter loaded cost: ~$0.75/min ($90K loaded ÷ ~2,000 hrs), derived
• Scheduling & coordination: roughly 35% of recruiter time goes to scheduling alone, per SHRM research

One number does most of the persuading. SHRM Foundation chair-elect Edie Goldberg estimates that 60% of hiring cost is soft cost: the time leaders and managers spend screening, scheduling, and making final decisions. The interviewing line is the heart of that 60%, which is why this model prices recruiter minutes, not job-board spend.

Worked example, 2,000-hire employer. 2,000 hires × 7 screens × 25 minutes × $0.75/min = 350,000 recruiter minutes, roughly $262,500 a year in screening labor. That's before a single unfilled seat. Factor in that roughly 35% of a recruiter's day goes to scheduling, and it climbs fast.

One clarification on vacancy cost, because this is where models usually overclaim. Don't count the whole cost of vacancy. Count only the days your interviewing process adds. The standard formula: annual revenue per employee ÷ working days × days vacant. For a revenue-critical technical role, standard cost-of-vacancy models put lost output in the thousands per month. If your screen-to-schedule gap adds five days to every fill, five days is what goes in the box. That discipline keeps the model credible under stress-testing. For sector-specific timelines, see Humanly's time-to-hire benchmarks by industry. Funnel-wide costs belong in the platform-wide ROI model.

Executive takeaway: Price the friction first. When the status quo has a dollar figure, the software stops looking like new spend and starts looking like cost recovery.

What manual phone screens actually cost

A manual phone screen runs $20 to $23 per completed screen once you count recruiter time, coordination, and no-shows. This is where interviewing automation pays for itself first. For modality-level pricing, see the automated phone screening buyer's guide. Here, the point is the formula so finance can rebuild the number from your inputs.

Here's the unit-cost box:

Cost per manual phone screen (minutes per screen × recruiter loaded cost per minute) ÷ completion rate Example: (25 min × $0.75) ÷ 0.80 completion = ~$23 per completed screen

Completion rate is the input most teams overlook. If only 80% of scheduled screens happen, every completed screen absorbs the cost of the no-shows around it. Drop to 50% and your cost per screen nearly doubles.

Scale it: 2,000 hires × 7 screens = 14,000 completed screens a year. At $20 to $23 each, that's $280,000 to $322,000 annually. That's your number for finance: the exact line automation collapses.

Humanly's platform data shows automated screening (SMS, chat, voice) at $2 to $4 per completed screen at scale, with completion rates near 80%. Run both through the formula above. The gap between $20 and $4 is the business case.

Executive takeaway: Cost per completed screen, not cost per screen, is the honest unit. Completion rate is the hidden multiplier, and automation fixes it directly.

Build the AI interviewing ROI model finance can check

The ROI on the interviewing line comes from four benefits, each tied to costs you already priced: recruiter minutes reclaimed, no-show losses recovered through higher completion, vacancy days saved by collapsing the screen-to-schedule gap, and fewer redundant rounds.

Interviewing ROI model (annual) recruiter minutes reclaimed × loaded cost per minute + no-show losses recovered (empty slots converted to completed screens) + vacancy days saved by the faster screen-to-schedule step × daily vacancy cost − annual software + implementation cost = net annual benefit Payback (months) = annual software + implementation cost ÷ net monthly benefit

Humanly-benchmark defaults for the benefits side:

• Time saved per early-stage screen: 25 minutes per candidate
• Time-to-fill reduction: ~11 days faster for high-volume roles
• Cost-per-hire reduction: ~17% lower, driven primarily by fewer recruiter hours
• Offer acceptance lift: 12% more accepted offers
• Recruiter throughput: 35–40% more candidates per week without adding headcount

Those are Humanly benchmarks. For independent evidence: a 2025 field experiment covering ~70,000 job applications (University of Chicago Booth and Erasmus University Rotterdam) found AI-led interviews increased job offers by 12% versus human-led screens, with higher job-start and 30-day retention rates. It's the largest controlled trial of AI interviewing published to date.

Worked example. Back to the 2,000-hire employer paying ~$262,500 in screening labor. Move 60% of screens to automation: roughly $150,000 in recovered recruiter capacity, plus a conservative $100,000 in recovered no-show losses and vacancy-day savings. Net benefit before software: ~$250,000. Platform plus implementation at $120,000 in year one puts payback inside six months.

Executive takeaway: Tie every benefit to a status-quo cost you already named. A benefit that doesn't map to a line finance recognizes won't survive the meeting.

The five questions your CFO will ask

Your CFO won't argue with the vision. They'll pressure-test the downside. Here are the five questions to answer in finance's vocabulary before they're asked.

1. "What if adoption lags?" Hand over the sensitivity table before they ask for it. Model payback at three capture levels so finance sees the floor, not just the ceiling.

• 100% benefit capture: ~$250K net annual benefit, ~6 months payback
• 75% benefit capture: ~$187K net annual benefit, ~8 months payback
• 50% benefit capture: ~$125K net annual benefit, ~11–12 months payback

Even at 50% of plan, payback stays inside a year. That's the number that wins the room.

2. "What's our compliance exposure?" Tools that score or rank candidates count as Automated Employment Decision Tools (AEDTs) under New York City's Local Law 144: independent bias audit, candidate notice at least 10 business days out, published summary of results, with penalties of $500 for a first violation and $500 to $1,500 for each subsequent one. Each day of noncompliant use counts as a separate violation, and similar rules are spreading. Pick a tool with a built-in bias-audit layer so compliance is a workflow property, not a fire drill. For the full framework, see Humanly's AI recruiting compliance playbook.

3. "What does implementation really cost?" Three buckets: integration, change management, and training. Look for native ATS integrations (Workday, iCIMS) so setup is configuration, not a custom project. Put a real number on change management too, because a tool recruiters route around returns zero.

4. "Why not just hire another recruiter?" Run the unit economics. Another recruiter adds one person's capacity at ~$90K fully loaded, and they'll still spend roughly a third of the week on scheduling. Automation adds capacity across the whole team and removes the admin instead of adding headcount to absorb it.

5. "What if we churn the vendor?" Ask about data portability. Can you export candidate records, transcripts, scores, and audit logs in a standard format? If evaluation data is trapped, you've bought lock-in, not leverage.

Executive takeaway: The CFO buys the downside case, not the upside. Walk in with the 50% sensitivity row, the compliance answer, and the exit plan already written.

The one-page executive summary your committee can forward

The artifact that survives the buying committee is one page finance can forward without you. Fill in every field and keep it to a single page.

AI Interviewing Business Case: One-Page Summary Problem: [screens/year, cost per completed screen, scheduling drag, no-show rate] Annual cost of manual interviewing: [$ from Section 1 + Section 2] Proposed spend: [platform + implementation, year one] Projected savings: [net annual benefit from Section 3] Payback: [months, with 50% downside row] Risks & mitigations: [adoption: phased rollout; compliance: bias audit; churn: data export]

Then sequence the room. Brief finance beforehand so the model isn't a surprise. Give your champion the sensitivity table. Pre-wire legal so nobody raises Local Law 144 mid-pitch.

Executive takeaway: The one-pager is the deliverable. If it can't be forwarded and understood without you in the room, it isn't finished.

Where Humanly attacks the model

Humanly targets the exact line items you priced above. The AI Recruiter screens candidates across chat, SMS, voice, and video, scores responses against your rubric, and books qualified candidates in the same conversation, collapsing the qualify-to-schedule gap from days to under two hours.

The Screen module saves around 25 minutes per early-stage screen, cuts about 11 days from time-to-fill, and lifts accepted offers by 12% versus human-led screens. Built-in bias-audit layer, native ATS integrations.

Want to run this against your own numbers? Book a 30-minute ROI working session. Bring your screens-per-hire and completion rate; leave with a payback period.

FAQs

How do you build a business case for AI interviewing?

Price the manual interviewing line first: hires per year × screens per hire × minutes per screen × recruiter loaded cost, plus scheduling drag, no-show losses, and vacancy days from the screen-to-schedule gap. Model the benefits (recruiter minutes reclaimed, no-show losses recovered, vacancy days saved), subtract software cost, and divide to get a payback period. Show a sensitivity table at 50%, 75%, and 100% capture so finance sees the downside case.

What do manual phone screens actually cost?

$20 to $23 per completed screen once you factor in recruiter time, coordination, and no-shows. Formula: (minutes per screen × recruiter loaded cost per minute) ÷ completion rate. At 2,000 hires and seven screens each, that's $280,000 to $322,000 a year. Automated screening runs $2 to $4 per completed screen at scale.

How do you calculate cost of vacancy?

A common formula: annual revenue per employee ÷ working days × vacancy days. For an interviewing business case, only count the days your screen-to-schedule gap adds, not the whole vacancy. Every day the interviewing step adds is a day of vacancy cost. Collapse that gap and you cut it directly.

Is AI interviewing compliant with bias-audit laws?

It can be, if the tool is built for it. NYC's Local Law 144 requires an independent bias audit, candidate notice, and a published summary of results for any tool that scores or ranks candidates. Pick a platform with a built-in bias-audit layer and loop in legal before the buying decision.

What's a realistic payback period for AI screening automation?

For high-volume employers, payback typically lands inside six to twelve months. Even at 50% benefit capture, payback usually clears within a year because reclaimed recruiter time and displaced agency spend are large relative to platform cost. Show all three capture levels rather than leading with the best case.