How much does an AI receptionist cost for a home-service business?
A practical worksheet for estimating AI receptionist cost from real call volume, provider charges, setup work, quality review, and human fallback.
Field note
By Vault Zero
An AI receptionist for a home-service business does not have one honest flat price. The useful estimate is monthly fixed cost + call minutes × all-in usage rate + setup and integration work + ongoing review + human fallback. Build that estimate from your own calls before comparing vendors.
Published rates are only inputs. At the time of writing, Vapi lists a $0.05-per-minute hosting charge and passes through model-provider costs, while live-receptionist providers can sell bundles of receptionist minutes; for example, Ruby publishes plans from 50 to 500 minutes per month. Neither number tells you what a reliable, business-specific workflow will cost by itself.
Start with a 30-day call audit
Export a representative month from the phone system and separate calls by outcome. If the business is seasonal, use a busy month and a normal month.
| Input | What to measure | Why it changes cost |
|---|---|---|
| Offered calls | Every inbound call that reached the business number | Establishes total demand |
| Answered calls | Calls answered by staff or an existing service | Shows current coverage |
| Missed and abandoned calls | Include caller hangups before a person answers | Defines the likely first workflow |
| Connected minutes | Talk time, not just call count | Most services have a usage component |
| After-call work | Time spent writing notes, texting, or updating a CRM | A cheap call can create expensive cleanup |
| Transfer attempts | Successful, declined, no-answer, and failed transfers | Exposes fallback requirements |
| Intake completeness | Whether name, callback number, address, issue, and timing were captured | Measures useful outcomes instead of conversation length |
Do not multiply total calls by an average duration from a sales deck. Use the calls the new system will actually receive: perhaps missed calls only, after-hours only, or all overflow calls.
Use an all-in AI cost formula
For each vendor, calculate:
Monthly AI cost = platform fee + telephony + speech-to-text + language model + text-to-speech + phone numbers + integrations + monitoring + human fallback + amortized setup
Ask for one all-in estimate using your selected voice, model, phone carrier, average duration, and concurrency. Usage-based voice stacks can combine several providers, so a platform price is not necessarily the final per-minute price. Vapi's own calculator separates hosting, transport, speech-to-text, language-model, and text-to-speech inputs, which is a useful checklist even if you choose another platform.
Setup belongs in the calculation too. Include discovery, call-flow writing, transfer rules, calendar or CRM connections, test calls, staff training, and revisions. If setup costs $S and you expect to use the workflow for N months, use S / N as the monthly setup allocation rather than pretending setup is free.
Compare the alternatives on the same scope
Existing staff
Count the time employees spend answering calls and completing follow-up, plus the cost of missed coverage. Do not treat a salaried employee's time as zero.
For a market reference—not a quote for your business—the U.S. Bureau of Labor Statistics reports that the median receptionist wage was $17.90 per hour in May 2024. Wage is not total employer cost. The BLS Employer Costs for Employee Compensation program separately measures wages and benefits, and reported private-industry benefits averaging $14.01 per hour in March 2026. Use local pay, payroll taxes, benefits, equipment, management, training, and the hours of coverage you actually require.
Human answering service
Include the base plan, included units, overage rules, transfer time, after-call work, holiday or after-hours treatment, onboarding, and add-ons. Confirm whether billing rounds by call, minute, or partial minute and whether a wrong number or spam call consumes the plan.
AI receptionist
Include every infrastructure layer, plus the human work that remains. AI may be able to answer simultaneous calls, but it still needs approved source material, exception handling, quality review, and a person or queue for calls it should not complete.
For a deeper apples-to-apples worksheet, use the AI receptionist versus answering service cost comparison.
Calculate cost per useful intake
Cost per minute rewards short calls even when they fail. A better operating metric is:
Cost per useful intake = total monthly operating cost / complete, correctly routed intake records
A useful intake should have a definition before the pilot begins. For a typical home-service request, it might require:
- caller name and confirmed callback number;
- service address or service area;
- plain-language reason for the call;
- caller timing or availability;
- the correct urgency lane under the business's approved rules; and
- successful delivery to the right person, queue, or system.
Track transfer success and staff correction time separately. A low-cost summary that reaches the wrong technician is not a saving.
A worked estimate without fake pricing
Suppose a contractor's audit finds 300 eligible calls per month at an average of three connected minutes. That is 900 forecast minutes. The worksheet becomes:
| Cost line | Your input |
|---|---|
| Fixed platform and phone-number charges | $_____ |
| 900 × quoted all-in usage rate | $_____ |
| Calendar, CRM, or messaging integration | $_____ |
| Monthly quality review hours × loaded labor rate | $_____ |
| Human overflow or escalation coverage | $_____ |
| Setup cost / expected months of use | $_____ |
| Estimated monthly total | $_____ |
Run the same sheet at 50%, 100%, and 150% of forecast minutes. Then add a version where calls run one minute longer than expected. This exposes plans that look inexpensive only at a narrow usage assumption.
Questions an itemized quote should answer
- Which charges are fixed, per minute, per call, per message, or per successful action?
- Which speech, model, voice, and carrier options are included in the estimate?
- How are silence, hold time, transfers, spam, and partial minutes billed?
- What happens to the call if the calendar, CRM, model, carrier, or vendor is unavailable?
- Who owns prompt changes, regression tests, and call review after launch?
- What data is retained, for how long, and what does reduced retention cost?
- Are setup, number porting, extra phone lines, support, and cancellation included?
- Can the business export call records and structured intake data?
The buying rule
Choose the smallest workflow that can produce a measurable operational improvement. For many plumbing, HVAC, roofing, and electrical teams, that means missed-call, overflow, or after-hours intake with explicit escalation rules—not an immediate replacement for every front-desk task.
Use the 30-day AI receptionist pilot scorecard to test the assumptions, or review the broader AI receptionist workflow before requesting a quote.
Put one call path under pressure before changing the whole phone system.
Tell me where calls get missed, what a useful handoff contains, and which situations still need a person. I'll map the narrowest pilot that can answer those questions honestly.