What missed calls are actually costing you
When you're under a sink, the phone is unreachable. When you're driving between stops, picking up is dangerous. When you're walking a homeowner through a quote, answering another call is rude. So you let it go to voicemail. This is not a discipline problem — it's the nature of field service work.
67% of inbound calls to home service businesses go unanswered during active service hours. That number comes from call data across plumbing, HVAC, and water treatment businesses — the same pattern in every trade where the people who do the work are the same people who answer the phone.
The missed call itself is not the problem. The problem is what happens next. 85% of callers who don't reach you call someone else. Most of them book with whoever picks up. They're not waiting by the phone for a callback — they're trying to solve a problem, and the next listing is one tap away. The average window before a caller is already booked with a competitor is under two hours.
A homeowner notices hard water at 9pm. The dishwasher is spotting, the soap isn't lathering, they've been thinking about calling someone for three weeks and tonight is finally the moment. They call you. You're closed. They move to the next listing.
After-hours callers are your highest-intent leads. They've already decided to buy — they're just figuring out who from. Your business is closed 16 hours a day. An AI receptionist captures every one of those calls for roughly the cost of a tank of gas per month.
How AI receptionists work — not a phone tree
A phone tree routes callers into menus. "Press 1 for scheduling. Press 2 for billing." An AI receptionist has a real conversation.
When your line gets a call you can't answer, the AI picks up in under two seconds. It speaks naturally. The caller speaks back. The system understands what they're saying regardless of how they phrase it — "my water tastes metallic and there's this beeping sound" and "I think my softener's broken" both get triaged correctly.
"Thanks for calling [Your Business] — I can help you schedule a service call or answer questions about your water treatment system." The caller knows they've reached your business, not a generic answering service. It uses your name and handles the call on your behalf.
The AI listens and identifies the type of call — new customer, existing customer, emergency, general question — and routes the conversation accordingly. It doesn't just collect information in a fixed order; it adapts to what the caller says.
Name, address, system type, what they're experiencing, best callback number, preferred appointment times. The questions it asks are the ones you'd ask if you were on the phone yourself — not a generic form, but a conversation calibrated to your services.
For routine service calls, it finds an open slot and books it. The caller gets a confirmation text with appointment details. Emergencies get flagged immediately. Complex inquiries get logged for callback with full notes already written.
Every handled call produces a written summary: caller name, issue, address, what was scheduled, any flags. You pull out of a customer's driveway and instead of a stack of voicemails you don't have time to return, you have a list of booked appointments and call notes waiting for you.
What water treatment calls look like handled by AI
Generic AI receptionists are trained on broad home service scenarios. Trade-specific tools — and well-configured generic ones — handle the specific situations your callers actually have. Here's what the four most common call types look like in practice.
Caller wants to schedule a salt delivery or check when they're next due. AI captures their address, confirms service area, checks available delivery slots, and books it — or queues it for your route planning if you don't use live calendar booking for deliveries.
Caller is experiencing hard water symptoms and wants to know about options. AI asks: city water or well? What are they noticing — spots, scaling, dry skin, poor soap lather? House size? It collects the qualification data before the conversation ends, so you're walking into the callback with a complete lead profile.
Water suddenly hard again, error light on, strange noise, unit won't regenerate. AI walks through diagnostic questions — when did it start, any recent power outage, salt level, any error codes on the display — and books a service call with those notes attached.
"My softener is leaking and there's water on the floor." The AI identifies this as urgent, provides immediate guidance on locating the bypass valve, and escalates to your emergency contact — not just queues it for next Tuesday. This is the call where human-backed AI tools earn their premium.
The after-hours new lead scenario deserves its own mention. Someone calling at 10pm isn't panicking — they're shopping. The AI collects their information, explains that someone will follow up in the morning, and sends them a text confirming they're on your callback list. They feel taken care of. You wake up to a qualified lead with all the intake data already collected.
Smith.ai vs. Goodcall vs. AgentZap
Three options representing the main approaches: hybrid human-backed AI, pure AI with a free entry tier, and trade-specific AI purpose-built for home service businesses.
- Hybrid model: AI handles routine calls, real virtual receptionists back up complex or unusual interactions — the highest quality floor of any option here
- Deep integrations with ServiceTitan, Jobber, and HousecallPro — books directly into your FSM calendar without a separate step
- Strong emergency triage — calls requiring real judgment get a real person, not a fallback recording
- Detailed call summaries emailed and texted after every interaction, ready to forward to your CRM
- Bilingual (English and Spanish) available on most plans
- Most expensive option of the three — not a free-tier testing ground
- Usage-based pricing can escalate with higher inbound volume
- Requires more onboarding configuration than pure-AI tools, which means more setup time upfront
From ~$255/mo · Usage-based above minimum
Plans include a set number of calls or minutes per month. Smith.ai adjusts their tiers — verify current pricing before signing up.
- Free tier available — prove the concept before spending anything
- Fastest to set up of the three: connect your Google Business profile, configure your services, start handling calls the same day
- Clean dashboard for reviewing call transcripts and refining your script
- Handles the most common call types reliably — scheduling, service area questions, hours, basic FAQs
- Flat monthly pricing on paid tiers, no per-call billing
- No human backup — pure AI only, which means unusual or confrontational calls may not be handled as gracefully
- Less trade-specific knowledge out of the box — needs training on your water treatment terminology and services
- Emergency escalation is basic; not designed for urgent call routing
- Simpler FSM integrations than Smith.ai
Free tier · Paid from ~$49/mo
The free tier handles a limited call volume — enough to evaluate whether the approach works for your specific call types before upgrading.
- Purpose-built for trades including plumbing and water treatment — understands the terminology and call types without heavy configuration
- Pre-built triage logic for home service scenarios: knows the difference between a salt bridge and a resin issue, asks the right diagnostic follow-up questions
- Designed to integrate with home service FSM platforms from day one
- Better cost-to-capability ratio than Smith.ai if you don't need human backup
- No human backup option — pure AI
- Newer product with a shorter track record than Smith.ai
- Fewer documented third-party integrations than more established tools
From ~$99/mo · Verify at agentzap.ai
The trade-specific pre-training reduces setup work compared to Goodcall, which partially justifies the price difference over the free tier.
| Feature | Smith.ai | Goodcall | AgentZap |
|---|---|---|---|
| Human backup | Yes | No | No |
| Trade-specific knowledge | Generic | Generic | Built in |
| Free tier | No | Yes | No |
| Starting price | ~$255/mo | Free / ~$49/mo | ~$99/mo |
| FSM integrations | Excellent | Basic | Good |
| Emergency escalation | Excellent | Basic | Good |
| After-hours coverage | Yes | Yes | Yes |
| Setup complexity | Moderate | Very low | Low |
Setting it up in a week
The setup is simpler than it looks. You need three things: a phone number that can be forwarded, a calendar or booking link, and 45–60 minutes to configure your script.
Call forwarding: your first decision
Most AI receptionist services give you a dedicated number. You configure forwarding from your existing line to that number. Conditional forwarding — forward after a set number of rings — is what most businesses use. Your phone rings first. If you can pick up, great. If not, the AI takes it. Unconditional forwarding sends all calls straight to the AI — better if you're consistently in the field and prefer the AI to handle every call professionally rather than you sometimes answering mid-job.
On most mobile carriers, conditional call forwarding is enabled by dialing *71 followed by the forwarding number, and disabled with *73. Landlines vary — check with your carrier.
Your first-week checklist
Services you offer, service area (cities and ZIP codes), rough pricing ranges, business hours, and any services you specifically do not offer. This is the foundation of every conversation the AI has on your behalf.
Enable conditional forwarding, then call your own business number from a different phone and let it ring through. Confirm the AI answers correctly, introduces your business, and handles a basic test inquiry. Fix anything before it handles real calls.
If you use Google Calendar, most services integrate directly. If you use a FSM like Jobber or ServiceTitan, connect it so the AI can see available slots and book appointments without double-booking or requiring manual entry.
Specify what constitutes an emergency — active leak, no water, equipment failure — and give the system a number to call or text when it detects one. This could be your personal cell, an on-call tech, or a shared team number. Don't skip this step.
Read the first 10–15 real call transcripts. You'll find gaps — questions the AI should be asking but isn't, answers that don't reflect how you'd say it, edge cases it handled awkwardly. Most services let you update the script in under 30 minutes. Do this at the end of week one, then again after week two.
How much are missed calls costing you right now
The calculator below works from your actual call volume and conversion numbers. The defaults are conservative estimates for a small water treatment operation — adjust them to match how your business actually runs.
How many inbound calls do you get on a typical business day, and what percentage do you currently answer? Think about active field hours — not when you're at your desk.
What's a typical job worth, and how often does a new caller who actually reaches you turn into a booked job?
Bottom line
Of all the AI implementations covered on this site, this one has the lowest barrier to entry. No hardware. No complex FSM integration required to start. No new workflow for your field techs. You sign up, configure a script, forward your calls, and it's running.
If you're regularly in the field and your phone is regularly going to voicemail, you're leaving a consistent stream of jobs on the table. $80/mo to recover most of those is the most straightforward ROI calculation in this guide series.
Start with Goodcall's free tier to see how your specific call types get handled. Move to AgentZap once you've validated the concept and want trade-specific performance without the overhead of configuring a generic tool. Choose Smith.ai if your average job value is high enough that a single fumbled emergency call justifies the cost of having a human available for edge cases.
Problems #02 and #07 — missed calls during field hours and after-hours lead loss — are listed separately because they feel different. One happens at 2pm on a Tuesday, one at 9pm on a Saturday. But they're solved by exactly the same tool. An AI receptionist that handles your midday calls also handles your evening calls. One subscription, two problems closed.
A real person will read your message and write back. No auto-responder.