AI ISA vs Human ISA: The Cost Math for Real Estate Teams

The honest version

You came here for a number, so here it is. A human ISA costs $50,000 to $85,000 a year fully loaded. An AI ISA platform costs $100 to $500 a month, plus per-minute usage above the included pool. On pure first-touch math, an AI ISA for real estate teams wins at almost every lead volume above 100 a month. On closing math, a human still wins the conversations that matter.

Most teams that get this right run a hybrid: AI handles the first 60 seconds and the 2:47 AM inquiry, a human handles the listing appointment. The vendor pages telling you to fire your ISA are selling you the wrong frame. Decide which conversations belong to software and which belong to a person, and the cost question settles itself.

TL;DR.

Why the ISA role exists at all

The ISA seat was invented to fix one number: speed-to-lead. The data on that number is grim and well-documented.

The WAV Group Agent Responsiveness Study found 48% of buyer inquiries were never responded to at all, and the average response time across the responders was 917 minutes (over 15 hours). Roof AI's secret-shopper test of 74 top-ranked brokerages found 41% never responded to a website lead, and only 9% replied within 5 minutes. (More on the underlying numbers in our lead response time statistics post.)

That's the wound the ISA role was built to close. A team lead hires an ISA because nobody on the listing side has time to triage 80 portal leads a week, and the cost of a missed first touch is the entire deal. The question now is whether the wound closes faster, and cheaper, with software.

What a human ISA really costs

That $50K–$85K range is the loaded number a CFO would actually defend, and base salary is only the first line of it.

Here's the build-up most teams I work with land on:

• Base + commission: $40,000–$60,000/yr depending on market. ZipRecruiter's real estate ISA salary data puts typical base pay in the low-to-mid $40Ks nationally, with stronger markets reaching the $50Ks before commission.

• Payroll tax, benefits, PTO: add ~22%.

• CRM seat, dialer, scripts, training: $1,800–$3,600/yr.

• Recruiting and onboarding amortized: $4,000–$8,000/yr, because in the teams we audit the seat turns over about once a year.

• Ramp loss: an ISA hits full productivity around month 3. Each turnover event costs about two months of pipeline.

That lands a competent, retained ISA at $55K–$75K, and a high-performing one in expensive markets at $85K+. Two ISAs covering nights and weekends: double it.

And you still don't have coverage at 11pm on a Sunday, which is exactly when a meaningful share of online inquiries arrives.

What an AI ISA for real estate teams costs

The sticker is $100–$500/mo. The real number depends on minutes.

Under the hood, AI voice agents have a fairly transparent unit cost: retail per-minute pricing runs $0.05–$1.00, and managed platforms typically resell at $0.25–$0.50/min once you add the routing, CRM integration, and call recording. For named reference points from that same source: Aircall prices its AI voice agent at $0.49/min on plans up to 2,500 minutes, and infrastructure-layer platforms like Retell start around $0.07/min before telephony, integration, and management get layered on. The subscription buys you a pool of included minutes plus a configured agent; you pay overage above the pool.

At 500 leads a month, a 3-minute average qualification call, and roughly 60% of leads reachable (the same assumptions as the break-even section below), you're looking at about 900 talk minutes, plus dial attempts and voicemail drops on the unreachable 40%. Budget 1,100–1,400 total minutes, close to the top of a typical included pool. Net monthly cost: $400–$700, all in. (For the broader cost frame across agent types, see our per-agent cost breakdown.)

The NAR 2025 Technology Survey is worth pausing on: only 17% of REALTORS report AI having a significantly positive impact. Adoption is broad, measured impact is rare. Our read on the 17%: they wired the tool into a specific lane and measured a specific number.

What AI handles well (and what it should never touch)

This is where vendor pages lie by omission.

Where AI is genuinely strong:

• First touch in under 60 seconds, 24/7, on every lead simultaneously.

• Structured qualification: budget, timeline, financing status, area.

• Calendar booking with a live agent's availability.

• Long-tail nurture sequences across SMS, email, and voice.

• Re-engaging aged leads at scale.

The hybrid model most teams actually run

The model that survives contact with real lead flow looks like this: AI takes the first touch, qualifies against a written rubric, books the appointment when criteria are met, and warm-transfers anything outside the rubric to a human. The human owns the conversation from there.

{{DIAGRAM_1}}

This is the same wiring we use for open-house follow-up: the AI runs the first 24 hours, the human runs the next 24 days. The economics work because the AI absorbs the high-volume, low-judgment portion of the funnel. That is the part that was breaking ISAs and resetting ramp every time the seat turned over.

One practical note from our builds. The rubric matters more than the model. Write down what "qualified" means for your team, in plain criteria the AI can check, and the hand-off stays clean. Leave it vague and the AI will book appointments your agents resent.

Break-even math by lead volume

Three volume scenarios, all with the same assumptions: 3-minute average qualification call, 60% of leads reachable, hybrid model where human covers ~20% of conversations.

Questions to ask any AI ISA vendor before signing

The vendor SERP is, charitably, a tire fire. Five questions cut through it:

1. What's the per-minute rate above the included pool, and where is it published? If they won't put it in writing, the overage will surprise you.

2. What's the escalation latency from AI to human? Sub-30 seconds is the bar. Anything slower and you'll lose hand-offs.

3. Do I get raw transcript and recording access, or only summaries? You need raw. Summaries hide failure modes.

4. What's the contract exit clause? Month-to-month or annual with a 30-day out is fine. 12-month lock-in with no exit is a flag.

5. What's the failure rate on calendar bookings, and how is it measured? A vendor that can't answer this hasn't measured it.

If a vendor stumbles on more than two of these, walk. These five are the first page of the audit we run. (For how this fits the rest of the stack, the operator's playbook for listing agents maps the full lane structure.)

FAQ

Can an AI ISA replace a human ISA on a real estate team?

Not for the conversations that close. AI handles first touch, qualification, and booking. That is the high-volume, rule-based part of the ISA job. The conversations that determine whether you win the listing still belong to a human. Teams that try full replacement usually rebuild the human role within months.

How much does an AI ISA cost per month?

Real-estate AI ISA platforms run $100–$500/mo base for the subscription, plus $0.25–$0.50/min for usage above the included pool. A typical mid-volume team at 500 leads/mo lands at $400–$700/mo all-in. Setup fees of $500–$3,000 one-time are common for script configuration and CRM integration.

What does a human ISA cost a real estate team per year?

Fully loaded, $50,000–$85,000/yr. That includes $40K–$60K base plus commission, ~22% payroll tax and benefits, $1,800–$3,600 in tooling, and amortized recruiting and ramp costs. In the teams we audit, the seat turns over about once a year, and covering nights with a second ISA doubles the bill.

What should an AI ISA never handle?

Pricing objections from sellers, grief/divorce/estate or relocation-under-duress conversations, repeat clients, referrals, and anything past the qualification hand-off. These are conversations where tone determines outcome and the wrong word loses the listing. Route them to a human on first contact and don't apologize for it.

What lead volume justifies an AI ISA?

The break-even line for the first-touch lane sits around 80–120 leads/mo. Below that, your listing agents can absorb the volume. Above that, an AI ISA for real estate pays for itself in the first billing cycle. At 500+ leads/mo, the hybrid model (AI first-touch, human close) beats both single-mode approaches on cost and coverage.

How should a small team prioritize real estate ai agent?

Start with the workflow that already has a baseline: hours, leads, errors, or budget waste.

What should be measured before investing in real estate ai agent?

Measure cycle time, volume, handoffs, error rate, and the current owner.

When should ai isa vs human isa cost math stay manual instead of automated?

Keep it manual when judgment, approval, brand nuance, or customer trust is on the line.

How does ai agent for real estate change the budget for real estate ai agent?

ai agent for real estate usually adds integration, QA, and monitoring work.

What is the first project to launch from this ai isa vs human isa cost math playbook?

Launch the narrowest workflow with a visible result.

Related reading

• Ads in Google AI Mode: What's Actually Buyable in 2026

• AI Email Personalization: What Lifts Revenue and What's Just Merge Tags

• AI for a Landscaping Business: Five Automations for About $345 a Month

• AI for Listing Agents: The Operator's Playbook

The actual decision

Pull your last 90 days of lead volume. Divide by three. Run it against the three columns above. If you're past the crossover line and still paying for human-only coverage, the math has already decided for you.

Book the audit and bring the spreadsheet. Thirty minutes against your actual lead flow will settle it.