How to Calculate ROI on an AI Voice Agent for Service Businesses

Every agency will tell you that their AI voice agent “pays for itself.” Almost none of them will show you the math.

That’s a problem, because the math either works or it doesn’t — and if it doesn’t work for your specific business, you shouldn’t buy a voice agent, no matter how good the pitch sounds. I’d rather lose a deal than have a client implement something that doesn’t deliver a return.

This article gives you the actual framework. Real formulas, real benchmark numbers from service businesses, and an honest look at the assumptions that have to hold for the math to work. By the end, you’ll be able to run this calculation for your own business in about 15 minutes.

The Core ROI Logic

At the most basic level, a voice agent ROI calculation is:

Monthly ROI = (Revenue recovered from captured calls) − (Monthly AI cost)

That looks simple. The complexity is in accurately estimating “revenue recovered,” because that number depends on three variables that most businesses don’t track closely:

1. How many calls are you currently missing or losing to voicemail?
2. What percentage of those missed calls would convert to paying jobs?
3. What’s the average value of those jobs?

Get those three numbers right, and the rest follows. Let’s build each one.

Step 1: Measure Your Current Call Loss

Pull your inbound call data for the last 30-60 days. Most business phone systems (and even call tracking software like CallRail or RingCentral) will show you:

• Total inbound calls
• Answered calls
• Missed calls / voicemails
• After-hours calls

If you don’t have call tracking software, your cell carrier and business phone system probably have this data in the settings. Go find it before you make any decisions.

What Typical Miss Rates Look Like

Based on what we observe when auditing call data for service businesses before building their voice systems:

• Roofing contractors: 30–45% of inbound calls missed during peak hours; near 100% after hours without an answering service
• HVAC companies: 20–35% missed during summer/winter peaks
• Plumbing and water damage: 15–25% missed during normal periods; much higher during weather events
• Dental offices: 20–30% missed during lunch, end-of-day, and weekends
• Law firms: 25–40% missed after hours — and those after-hours calls are often the highest-urgency leads

If your miss rate is under 10% during business hours, a voice agent’s inbound capture value is limited. If you’re above 20%, there’s almost certainly money being left on the table.

After-Hours Is Different Math

After-hours missed calls deserve special attention. These aren’t just calls that went to voicemail while your office manager was handling another line. These are calls that received zero response until the next business day — if ever.

The lead decay rate for service businesses is steep. Studies on speed-to-lead consistently show that a lead contacted within 5 minutes of inquiry converts at 3–5x the rate of a lead contacted after an hour. A lead that waits until the next morning is almost certainly calling your competitor at 8:05 AM.

An after-hours call that goes to voicemail has, in practice, a near-zero conversion rate. An after-hours call handled by a voice agent has a conversion rate that approaches your normal business-hours rate — because the lead is qualified and booked before they have a chance to shop around.

Step 2: Estimate Your Missed Conversion Rate

Not every missed call would have converted. Some are existing customers with simple questions. Some are price shoppers who aren’t serious. Some are wrong numbers.

When calculating ROI, use a conservative conversion rate — the percentage of your missed calls that would realistically become paying customers.

Industry Benchmarks for Inbound Call Conversion

These are ranges we’ve observed from service business call data:

Industry	Avg. Inbound Call Conversion Rate
Roofing (storm damage, urgent)	18–30%
Roofing (planned jobs)	10–18%
HVAC (emergency service)	25–40%
HVAC (planned maintenance)	12–20%
Plumbing (emergency)	30–50%
Dental (new patient)	20–35%
Personal injury law	8–15%
General home services	12–22%

Use the lower end of your industry’s range when modeling ROI. If reality outperforms your conservative estimate, that’s a pleasant surprise — not a problem.

Step 3: Know Your Average Job Value

This is usually the easiest number to pull from your accounting software. Calculate the average revenue per job over the last 12 months. Include all job types (repairs, replacements, inspections) to get a true average.

For jobs with high variance (a roofing job can be $1,500 for a repair or $18,000 for a full replacement), it’s worth calculating separately for:

• Emergency/urgent jobs — typically higher average value, also highest probability callers are the ones ringing you at 8 PM
• Planned/quote jobs — lower urgency, lower immediate conversion probability

For the ROI calculation, use your overall average job value unless you have solid data showing that after-hours calls or missed calls skew toward a particular job type.

The Full ROI Formula

Here’s the complete monthly calculation:

Missed calls per month
× Realistic conversion rate (conservative estimate)
× Average job value
= Gross revenue recovered

Gross revenue recovered − Monthly voice agent cost = Monthly ROI

And the payback period on any upfront build cost:

Upfront build cost ÷ Monthly ROI = Payback period in months

A Real-World Example

Let me walk through an example based on a roofing contractor we’ve worked with — I’m keeping the name and location generic, but these are real parameters.

Their situation:

• 180 inbound calls per month
• 52 missed calls per month (29% miss rate — typical for a 4-person operation)
• Average job value: $7,200
• Previous conversion rate on answered calls: 22%

Voice agent ROI calculation:

Calls the voice agent captures per month: 52 missed × 80% capture rate = 41.6 calls (agents don’t capture 100% — some callers hang up before the agent completes the greeting)

Converted jobs from captured calls: 41.6 × 22% conversion = 9.2 additional jobs per month

Revenue recovered: 9.2 × $7,200 = $66,240/month

Monthly voice agent cost: $1,000/month subscription

Monthly ROI: $65,240

Even cutting the conversion rate in half to 11% (extreme pessimism), the math produces $32,000+ per month against a $1,000 investment.

The payback on a custom build ($15,000 upfront): less than one month.

This is why voice agents make obvious financial sense for call-heavy service businesses. The math gets harder to dismiss when you’re looking at it from the perspective of what each individual missed call is actually worth.

The Assumptions That Can Break the Math

I don’t want to oversell this. There are scenarios where the ROI math doesn’t work, and you should understand them.

Low Call Volume

If you’re receiving 15 calls per day total and your miss rate is 8%, there’s no significant call capture problem to solve. An AI voice agent adds cost without meaningful impact. The math requires a meaningful volume of missed calls to work.

Very Low Average Job Value

For a business where the average transaction is $80 and most inquiries are existing customers asking operational questions, the revenue recovery math is thin. Voice agents shine most where individual deals are worth $1,000+.

High Answering Service Coverage

If you already use a human answering service that captures and qualifies calls outside business hours, your uncaptured call volume may already be low. In this case, the ROI comes from replacing an expensive human service ($800-$1,500/month for quality 24/7 coverage) with a better-performing AI alternative.

Poor Lead Quality in Your Market

In some markets and industries, a high percentage of inbound calls are unqualified. If you’re in a high-SPAM-call environment or your marketing attracts a lot of tire-kickers, your realistic conversion rate on captured calls will be lower than industry benchmarks.

Run your own numbers. Don’t use generic benchmarks as gospel for your specific business.

The Costs You Need to Include

Make sure your cost side of the equation is complete. Voice agent costs come in layers, and the full picture is different from the headline price.

Voice Agent Subscription Model

If you go with a subscription (like our $1,000/month offering), the monthly cost is all-in — build, platform, API usage, and ongoing maintenance. This is the simpler cost model and the one that makes ROI calculation straightforward.

Custom Build Model

Upfront build cost ($8,000–$25,000) plus ongoing monthly costs:

• Platform and hosting: $150–$500/month
• API usage (speech-to-text, LLM, text-to-speech, telephony): $200–$800/month at 300–500 calls/month
• Maintenance and optimization: $300–$800/month

Total ongoing after custom build: $650–$2,100/month

The custom build makes financial sense when the ongoing monthly cost savings versus subscription are significant over a 24-36 month horizon, or when you need a level of customization that subscriptions don’t offer.

The API Cost Surprise

This catches businesses off guard: API costs scale with volume. If your voice agent handles 200 calls in month one and 600 in month six (maybe you added a Google Ads campaign, maybe storm season hit), your API bill roughly triples. This is fine — more calls means more revenue — but budget for it. Model API costs at 2x and 3x expected volume.

At a rough estimate, a 5-minute voice call costs $0.25–$0.75 in combined API costs (speech recognition, LLM processing, voice synthesis, telephony). At 500 calls/month averaging 4 minutes: $500–$1,500/month in API costs alone.

For a deeper comparison of what these costs look like across the real cost categories involved, see our post on the real cost of AI implementation for small businesses.

Calculating the Outbound Component

Everything above assumes inbound call capture. But voice agents can also run outbound — calling old leads, reactivating dormant customers, confirming appointments.

The ROI math for outbound is different:

Outbound campaign cost = Contact list preparation + AI call costs + Human follow-up time

Outbound campaign return = (Leads contacted) × (Re-engagement rate) × (Conversion rate) × (Average job value)

For roofing and other seasonal service businesses, outbound to the previous year’s estimate list at the start of a new season is one of the highest-ROI uses of a voice agent. We’ve seen reactivation campaigns convert 4–8% of cold estimate lists into booked jobs — at near-zero additional marketing cost.

Building Your Own Spreadsheet

Here’s the simple version you can build in 5 minutes:

Input	Your Number
Monthly inbound calls
Current miss rate (%)
Missed calls per month	(auto-calculate)
Voice agent capture rate	75-85%
Calls captured per month	(auto-calculate)
Conversion rate (%)
Additional jobs per month	(auto-calculate)
Average job value ($)
Gross revenue recovered ($)	(auto-calculate)
Monthly voice agent cost ($)
Monthly ROI ($)	(auto-calculate)

Fill in the first four manual inputs and the rest flows. If the monthly ROI number is less than 3x your investment, dig into why — usually it’s either a low miss rate (the capture problem is smaller than expected) or a very low average job value (each recovered call doesn’t move the needle enough).

When to Use This Framework for Your Decision

Run this calculation before any sales call with an AI vendor. Walk into that conversation already knowing your numbers — your call volume, your miss rate, your job value. It immediately filters out vendors who try to sell you on generic “massive ROI” claims without looking at your specific data.

Any honest AI agency should be able to look at your numbers and tell you whether the math works. If they can’t do that, or if they don’t ask for your call data before making ROI claims, that’s a red flag worth paying attention to.

The math either works or it doesn’t. Do the calculation, and the answer will tell you what to do.

Frequently Asked Questions

What’s the minimum call volume where a voice agent makes financial sense?

As a rough rule of thumb: if you’re receiving fewer than 50 inbound calls per month, a voice agent is hard to justify on pure ROI. Below that volume, the number of missed calls is typically too small for the revenue recovery math to outpace the cost. For businesses getting 100+ calls per month with a miss rate above 20%, the economics are usually favorable within the first month.

How do I accurately measure my current miss rate?

The most reliable method is call tracking software (CallRail, WhatConverts, RingCentral Analytics). If you don’t have that, your business phone provider often has call logs available. Pull data for at least 30 days — ideally 60-90 to account for seasonal variation. Count calls that went to voicemail, were missed entirely, or rang without being answered. Divide by total inbound calls. That’s your miss rate.

Can I calculate ROI for an outbound voice agent separately?

Yes, and the math is actually simpler. Take your existing lead or customer list, estimate a realistic re-engagement rate (3-8% is typical for warmer lists), apply your conversion rate, and multiply by average job value. Compare that to the cost of the outbound campaign (API costs + list prep time). For service businesses with a dormant estimate list or lapsed customer database, outbound campaigns often show positive ROI within the first campaign — especially before a seasonal push.

Does the voice agent ROI calculation change if I already have an answering service?

Yes. If you already pay $800-$1,500/month for a human answering service, the cost comparison changes. You’re not comparing “AI cost vs. nothing” — you’re comparing “AI cost vs. current answering service cost + AI performance differential.” In most cases we’ve seen, AI voice agents outperform human answering services on conversion rate (because the AI is better at qualification and booking) while costing the same or less. The net ROI improves compared to the baseline calculation.

How long until a voice agent investment pays back?

For a subscription model ($1,000/month), there’s no upfront investment to recover — the ROI is positive from month one if the math holds. For a custom build ($15,000-$25,000), payback period depends on monthly ROI. At $5,000/month net benefit (conservative for a mid-size service business), payback takes 3-5 months. At $15,000+/month net benefit (realistic for a roofing company during storm season), payback can happen in the first month. The subscription model removes payback period risk, which is why we structured our offering that way.