Every HVAC and plumbing owner already knows the script. The phone rings at 11:42 PM on a Sunday in August. The customer’s upstairs AC died. They try you first because your truck wrapped their neighbor’s driveway last week. You don’t answer. They Google your nearest competitor. The job leaves your office before the homeowner finishes their cup of water.
That single missed call was worth between $640 and $4,200, depending on whether it became a service call, a new system install, or a five-year maintenance customer. Multiply that by the after-hours calls you missed last August, and you understand exactly why your top competitor is taller this year than they were last year.
This playbook is the engineering blueprint for never missing one again. It is the same operating system Nirvani has deployed across 41 home services companies (HVAC, plumbing, electrical) and the average outcome is documented at the bottom. We’re going to walk through five connected modules. The voice agent is the spine. The other four modules are the limbs that pull money out of customers you already have.
No fluff. No marketing-speak. If you run a residential or light-commercial home services shop with one truck or twenty, this is the architecture. The order matters. The scripts matter. The integrations matter. Read it once. Bookmark it. Build it.
You are losing real money on the phones.
This isn’t a soft problem. It isn’t about branding or vibe. There are three brutal numbers that describe what is happening in your office every single week. They come from Nirvani deployment data across 41 home services companies (HVAC, plumbing, electrical), cross-referenced with industry benchmarking from BrightLocal, Service Roundtable, and ACCA member surveys.
Look at those three numbers together. Roughly one in three of your urgent inbound calls leak. Of the ones you do catch, fewer than three in twenty turn into a recurring revenue customer. And of the ones who do get on your schedule, your dispatch window is so wide that the homeowner has time to call a second shop and start a bidding war while waiting on hold.
Every owner reading this is going to say my numbers aren’t that bad. That’s what every owner says. The 41 shops we’ve audited all said it too. Then we listened to one week of their inbound call recordings and pulled their CSR scoreboards from their field service management system. The numbers were that bad. Sometimes worse.
The good news is the fix is mechanical. It’s a stack, not a vibe shift. You install the modules in order. The compounding starts the day after the voice agent goes live.
What the stack delivers.
These are not aspirational numbers. They are the average outcomes across the 41 home services deployments after 90 days of running the full stack. Some shops do better. The minimum performance gate to be counted is 60 days post go-live with all five modules in production.
Pay attention to the dispatch number. We are not promising a 47-minute on-site arrival. We are promising 47 minutes from the inbound call to a confirmed tech, named, in the customer’s phone, with a real ETA and a real truck rolling. That is the moment the homeowner stops shopping.
The 7-week payback figure assumes you are running on Nirvani N1 at $444/mo. If you’re running on N2 at $1500/mo (the tier that includes dedicated CSM and the field service management integration suite), the payback figure averages closer to 9 weeks. After payback, every additional emergency captured and every additional maintenance plan attached drops to bottom line.
The five-module Revenue Stack.
Each module solves one revenue problem. They share one customer record, one phone number, one calendar, and one SMS thread. Together they form a closed loop from first ring to post-service review.
Order matters. Module 01 is the spine and goes live first. Modules 02 through 05 layer on in roughly the order shown. You can technically launch them in parallel if you have the operational bandwidth, but most shops walk it in week by week so the CSRs and techs can adapt without burning out.
The job. Answer every inbound call in under one second, day or night, including the 11:42 PM Sunday calls. Triage for safety. Determine if it is a true emergency, an urgent same-day, or a routine. Capture the address, the system type, and the homeowner’s name. Book the dispatch slot directly into your field service management calendar. Send the homeowner a confirmation SMS with the tech name and ETA window inside 90 seconds.
Why this is module 01. Every other module assumes the call got answered. If you don’t solve the answer problem first, the rest of the stack is decoration. This is the structural change. Everything else is paint.
Sample script: After-hours emergency call
This is the verbatim script template Nirvani deploys for residential HVAC and plumbing emergencies. Tone is calm, fast, urgency-aware. The first three turns determine whether the customer feels safe.
Integration points
The voice agent reads from and writes to your field service management system. Whether that’s a category leader like ServiceTitan, a mid-market platform like Housecall Pro, an enterprise system like FieldEdge, or a workhorse like Jobber, the agent integrates at the dispatch API layer so it can see live tech availability, write new jobs, and pull customer history if the caller is recognized.
When the caller is an existing customer with an active maintenance plan, the script tone shifts. The agent pulls the customer’s name, last service date, the system make and model (SEER rating, BTU capacity, refrigerant type if it’s R-410A or R-454B), and any open warranty work. The dispatch quote includes their plan-member rate, not the street rate. Plan members feel the difference on the second call. That’s how you keep them.
What you configure once
- Greeting. Your business name. Your CSR persona name (we recommend a real human first name. Maya, Daniel, Sofia. Avoid robotic names.).
- Service area. Zip codes you cover. Out-of-area handling (decline gracefully and offer referral).
- Emergency definition. What counts as emergency for your shop. Most HVAC shops define emergency as no cool below 85F outdoor temp, no heat below 35F outdoor temp, or active water intrusion. Plumbing shops define it as active flooding, no water, or sewage backup.
- On-call rotation. Which tech is primary on which nights. Backup escalation chain.
- After-hours pricing. Dispatch fee, first-hour rate, after-hours surcharge. Flat rate vs T&M.
- Disqualification rules. When the agent politely declines (commercial calls if you’re residential-only, calls outside your service area, warranty work routed to manufacturer).
What the agent does NOT do
Module 01 is not a diagnosis tool. It does not tell the customer what is wrong with their compressor or their water heater. It does not give a price for the repair. It does not commit to a fix. Its job is to triage, dispatch, and confirm. The tech does the rest. Trying to extend the voice agent into diagnosis is the single most common failure mode and we coach every shop away from it on day one.
The job. At the moment the tech presents the repair or install quote, automatically surface the maintenance plan option as an in-the-flow line item with the savings math pre-calculated. Make it impossible to forget. Make it impossible to skip. Make the math obvious.
Why it works. The single biggest reason attach rates sit below 15% isn’t price objection. It’s tech inconsistency. Some techs pitch it every time. Some never bring it up. The ones who do pitch it use ten different versions of the pitch and only three of them work. This module standardizes the pitch into the quote workflow so it gets delivered identically on every job, by every tech, every time.
Sample script: Tech-led upsell at quote close
Plan structure
Two tiers. Resist the temptation to build three. Three confuses, two converts. We have tested this on 41 deployments and the data is unambiguous.
- Basic tier ($14 to $19/mo). One annual tune-up, 10% repair discount, priority scheduling, reduced diagnostic fee. This is the volume tier. Aim for 40 to 55% of new customers attaching.
- Premium tier ($29 to $39/mo). Two annual tune-ups (one for cooling, one for heating), 15% repair discount, no after-hours surcharge, free dispatch, transferable to a new home, two-year price lock. This is the profit tier. Aim for 12 to 20% of new customers attaching.
Premium-tier members are roughly 4.5x more profitable annually than non-members and 2.3x more profitable than basic-tier members because they call you for everything, they call you first, and they never shop on price.
Objection handlers that actually work
- “I don’t want a contract.” “You can cancel anytime with one text. I’ll make a note in your file that you specifically asked about that.”
- “I have a home warranty.” “Home warranties don’t do tune-ups. They pay out on broken parts after long waits. The membership prevents the break in the first place.”
- “I’ll think about it.” “Totally fair. If you want, I can apply today’s 15% discount and you can cancel before the second month if it doesn’t feel worth it. Either way you get $111 off this ticket right now.”
- “My system is too old to bother.” “Old systems need it most. The membership pays for itself the day you need a callback. And if you replace, you keep the discount on the install.”
Tech scoreboard
Every tech sees their attach rate on the dispatch tablet. Weekly leaderboard goes to the team Slack or text thread. The top performer gets a $100 spiff. The bottom performer gets a 15-minute ridealong with the top performer the following week. Attach rates respond to visibility faster than they respond to incentives. Make the number visible and watch the team self-correct.
The job. Two weeks before the first 90F day of the year (cooling season) and two weeks before the first 35F night of the year (heating season), automatically send every customer who is due for a tune-up an SMS offering them a pre-season slot at a slight discount. Fill the shoulder weeks before the phone explodes. Smooth the demand curve.
Why it works. The customer who books their AC tune-up on April 15 doesn’t call you in July when their unit dies. The one who waits until July is the one who shops the emergency call to four competitors. Pre-booking is the cheapest insurance against demand collapse on the customer side and tech burnout on yours.
Sample script: Pre-season SMS
Trigger logic
- Cooling season trigger. First forecasted 90F day in the customer’s zip code 14 days out (NOAA forecast feed).
- Heating season trigger. First forecasted 35F overnight low in the customer’s zip code 14 days out.
- Plan member priority. Premium plan holders get the first wave (slots reserved). Basic plan holders get wave two. Non-members get wave three with the standard rate.
- Frequency caps. Two sequences per customer per year, max three messages per sequence, hard STOP/HELP compliance.
- Self-serve booking link. Reply with a time, or tap the link for a calendar picker. Both write directly to the field service management dispatch board.
What the data shows
Across the 41 deployments, shops running module 03 averaged a 34% increase in March-April tune-up bookings and a corresponding 22% decrease in June through August emergency call volume. The decrease in emergencies is the underrated win. It means your techs are not burning out in peak season. It means your CSRs are not on hold queues. It means your customer experience stays good when your competitor’s collapses.
The conversion math on the SMS sequence itself: average reply rate is 18%, average book-from-reply rate is 71%, net booking rate per send is ~12.8%. On a customer base of 2,400 active accounts, a single seasonal sequence books ~307 tune-ups at $89 each. That is $27,300 of pre-booked revenue per season per sequence. Two seasons per year. $54,600 added annually from module 03 alone.
The job. The morning of the appointment, send the customer an SMS confirming the visit and giving them a two-hour window. Thirty minutes before arrival, send a second SMS with the tech’s name, photo, truck number, and a live tracking link. The moment the tech taps “arriving” in the field service management app, send a third SMS with the actual ETA.
Why it works. The number-one driver of cancellations and reschedules in home services is not price. It is the customer waiting 4 hours for an open window. They miss a call from work, they get frustrated, they cancel. The two-hour window plus the tech-arriving confirmation kills cancellations cold. Across the 41 deploys, average cancellation rate dropped from 14% to 4% within 60 days of module 04 going live.
Sample script: Same-day confirmation sequence
What this costs the customer to do nothing about
The reschedule rate inside the field service management dispatch board hides a real number: when a customer cancels at 9:30 AM for a 10 AM appointment, your tech has 30 minutes of wasted drive time, your dispatcher has to scramble a backfill, and the slot usually goes unfilled. The hidden cost of every cancelled job ranges from $180 to $640 in lost slot value plus drive time. Module 04 turns that drag into a tailwind.
Reschedule self-serve
If the customer replies R, the system replies with three alternate slots within the next 48 hours pulled live from your dispatch board. The customer picks one with a single tap. No CSR involvement. No phone tag. Reschedule completed in 90 seconds.
The job. The moment the tech taps “job complete” in the field service management app, send the customer an SMS thank-you with two clear next steps. (1) A direct link to leave a Google review. (2) A referral code for $25 off their next service plus $25 to the friend they refer.
Why it works. Reviews compound. Every 0.1-star bump on Google moves your local pack ranking, your CTR, and your call volume. The shops that systematically ask for a review on every completed job average 4.8 stars with 1,200+ reviews after 18 months. The ones that don’t ask sit at 4.2 with 80 reviews. The first group dominates local search. The second group buys ads to compensate.
Sample script: Post-service handoff
The negative-review interception
This is the underrated feature. When a customer replies 1, 2, or 3, the system never sends them a public review link. Instead it sends them a private form that pings the owner’s phone directly with the rating, the customer name, and the tech’s name. The owner has roughly 90 minutes to call the customer back and resolve the issue before the customer cools down and writes the public bad review on their own. Across 41 deploys, this single mechanic reduced public 1-star and 2-star reviews by an average of 64%.
Cross-platform routing
Google is the primary destination because it drives local pack ranking. After 30 days, if the customer has been a 5-star multiple times, the system invites them to leave a review on a secondary platform (Angi for HVAC, Yelp for plumbing, BBB for both). This spreads the social proof across the platforms where buyers actually shop. No single platform dominates the funnel.
Referral economics
The $25 referrer credit, $25 friend discount structure works because it is symmetric. Asymmetric referral programs (where the referrer gets a lot and the friend gets nothing) underperform. The friend needs a real reason to click. Across the 41 deploys, the average customer who is asked for a review and given a referral code refers 0.42 new customers per year. On a customer base of 2,400, that’s ~1,000 referred customers per year. At an average lifetime value of $1,650 per referred customer, that’s $1.65M of referred lifetime revenue. The cost is $25 x 2 x conversion rate. Real-world net is roughly $42 of incentive cost per referral, scaled to about $42,000 per year against $1.65M of revenue generated.
30 / 60 / 90 Day Roadmap.
This is the standard rollout path. Roughly 80% of the shops we deploy follow it as-is. The other 20% compress to a 60-day rollout because they have a strong office manager and they want to move faster. We do not recommend compressing further than that. The team adoption curve matters more than the technology timeline.
01-30
Foundation
- Voice agent deployment (Module 01)
- Connect to your field service management system
- Configure greeting, service area, on-call rotation
- Load your pricing rate card
- Train CSRs on the new handoff workflow
- Pull baseline call metrics (week 1)
- Two CSR ride-alongs with the voice agent
- First 30-day reporting checkpoint
31-60
Conversion
- Maintenance plan upsell (Module 02) live
- Two-hour window confirmations (Module 04) live
- Tech tablet training on plan upsell scripts
- Tech scoreboard goes live
- Plan structure (basic + premium) priced
- Objection handler training, two sessions
- Reduce cancellation rate target: 14% to 6%
- Attach rate target: 25%+ by day 60
61-90
Compounding
- Seasonal pre-book automation (Module 03) configured
- Review + referral loop (Module 05) live
- Negative-review interception trained
- Plan member priority routing on
- First seasonal SMS sequence sent
- Google review velocity target: 8-12 / week
- Full-stack 90-day reporting + tuning
- Optional: N1 to N2 tier upgrade discussion
The two-week pre-deployment checklist
Before day 01, the office does this work. It takes about 6 to 10 hours total across two weeks. Doing this work up front is the difference between a clean go-live and a messy one.
- Inventory your call data. Pull 30 days of inbound call records from your phone system. Tally answered vs missed, after-hours vs business hours, by day of week. This becomes your baseline.
- Document your dispatch rules. Write down your on-call rotation, your emergency definition, your service-area zip codes, your decline rules, your after-hours pricing.
- Audit your pricing. Confirm your rate card in the field service management system is current. The voice agent reads from it. Stale prices mean stale quotes.
- Tag your customers. Make sure your existing customer records are tagged with plan membership status, system age, last service date, and preferred tech if any. The voice agent uses these tags to personalize.
- Pre-write your CSR handoff script. When the voice agent escalates a call to a CSR, what do they say first? Write it down. Practice it.
- Tech ridealong. Pick one veteran tech and one rookie. Both ride together for one full day with the voice agent in shadow mode. They surface every operational quirk the agent needs to know.
ROI snapshot.
Below is the average 12-month ROI snapshot across the 41 deployments, for a residential HVAC or plumbing company doing roughly $2.1M in annual revenue with a 6-truck fleet and 2,400 active customer records. Your numbers will differ. The directional lift is what to focus on.
12-Month Stack Performance
| Metric | Before | After | Lift |
|---|---|---|---|
| Emergency call capture rate | 68% | 95% | +27pp |
| Maintenance plan attach rate | 14% | 42% | +3.0x |
| Avg ticket value, all jobs | $612 | $784 | +28% |
| Maintenance plan revenue, annual | $58,000 | $210,000 | +$152K |
| Cancellation rate | 14.2% | 4.1% | -71% |
| Avg dispatch latency, emergency | 4h 18m | 47m | -82% |
| Google review velocity, weekly | 1.2 | 10.4 | +8.7x |
| Public 1 to 2 star reviews, annual | 23 | 8 | -65% |
| Referred new customers, annual | 68 | 412 | +6.1x |
| Net additional revenue, annual | $0 | $487,000 | 7wk payback |
Three things to point out. First. The maintenance plan revenue is recurring. Half a million in additional revenue is a meaningful number on its own. Recurring half a million is a category change to the business. Second. The cancellation drop converts directly to gross margin because the technician hours were already paid for. Third. The referred customer figure compounds. Year two of running the full stack averages another 15 to 20% above year one because the referral flywheel takes 9 to 12 months to fully spool up.
Case study: Cooper & Sons HVAC.
Composite of three real deployments. Numbers averaged. Names changed. Geography typical.
Cooper & Sons HVAC: 6 trucks, residential focus, Tempe Arizona
Family-owned HVAC and light plumbing shop. Father started the business in 1998, son took over in 2019. Six trucks, eight techs counting one apprentice, two CSRs, one dispatcher, one office manager. Annual revenue around $2.1M. Customer base around 2,400 active records, mostly residential single-family in the East Valley.
The pain
June 2025 was the breaking point. Phoenix hit 117F for nine days. Cooper’s after-hours line went to an answering service that mangled three of five dispatches. The fourth one (a $14,400 system replacement) walked to a competitor. The fifth one (a maintenance plan member with a fried capacitor) escalated to a 1-star review because the tech showed up at 8 AM after the customer called at 11 PM.
The deployment
Sammy and the Nirvani team kicked off in week one of August 2025. Voice agent live August 14. Maintenance upsell scripts live September 8. Window confirmations live September 12. Seasonal pre-book configured October 2 (ready for the heating-shoulder season in November). Review loop live October 15. Full stack operational at day 67.
The results
The owner’s comment at the 12-month review: “The maintenance plan revenue alone covers our two newest trucks. The voice agent is the thing my wife notices first because she stopped getting woken up by the on-call rotation drama.” Cooper & Sons is now adding a 7th truck and a second light-commercial division. They cite the stack as the operational enabler.