A 3-truck HVAC company was bleeding revenue from three holes at once — missed calls, slow callbacks, and a dead customer database — while paying more every month for leads it couldn't convert. We built one AI-driven engine that manufactures demand the moment the market is ready, answers and books every call within seconds, and turns one-time jobs into lifetime memberships. Cost per booked job fell from $310 to $156, and missed-call text-back alone recovered ~$40K a month in jobs that used to go straight to voicemail.
Figures are blended across the company's HVAC and plumbing service lines. The model generalizes to roofing, plumbing, electrical, garage doors, pest control, and solar.
Family-owned HVAC and plumbing, one metro, three trucks, ~$1.3M a year. Great technicians, no marketing. The owner was the marketing department — boosting a Facebook post between jobs and paying a vendor $61 for shared leads four other companies were also calling.
The money was leaking through three holes: ~38% of inbound calls went to voicemail (a dead lead in home services), captured leads sat in a notebook for 3.5 hours until someone called back, and ~6,000 past customers were never contacted again. We didn't sell them ads — we built an operating system for the whole revenue loop.
One dashboard running a continuous Trigger → Capture → Convert → Maximize → Prove cycle. Each part feeds the next, and completed-job revenue data flows back to the top — so the engine buys demand it can actually service and optimizes for booked customers, not cheap clicks.
Home-services demand is triggerable, not random. The Radar watches weather, seasonality, and local signals — pre-positioning spend ahead of a heat wave or hard freeze instead of a day late. It's wired to the dispatch board, so it throttles ad spend down when the schedule is full and pushes it up when there are open truck-hours, and bids harder for high-margin work like replacements than for $89 tune-ups.
When an AC dies, the homeowner Googles, taps the map pack, and maybe asks Facebook — all in 90 seconds. We unified Local Services Ads, Search, Google Business Profile, and Meta into one orchestrated front, with a retargeting mesh that follows anyone who calls-but-doesn't-book until they convert. Budget flows toward whatever produces profitable jobs, not cheap clicks.
You can't grow a bucket with a hole in it. Any unanswered call gets an instant text-back in seconds, a 24/7 AI receptionist books overflow and after-hours calls straight onto the dispatch board at 2am, and homeowners can text a photo of the unit for a ballpark and a booking. Answered-and-recovered call rate went from ~62% to ~98%.
Reviews drive map-pack rank, paid conversion, and close rate all at once. The moment a job is marked complete, the customer gets a one-tap review link at the peak 'it works again' moment. Unhappy signals route to a manager for a private save first, and AI replies to 100% of reviews. The flywheel: more reviews → higher rank → more free calls → lower blended cost-per-lead.
Every lead is qualified and booked in real time — 22-second average speed-to-lead — then scheduled in geographic clusters to cut windshield time and fit more jobs per truck per day. Automated confirmations and 'tech is on the way' sequences crush no-shows, and nearby past customers get a soft 'we'll be in your neighborhood Tuesday' offer to fill dead time with high-margin work.
This is where the real money was. After every job, an automated sequence converts one-time customers into recurring maintenance-plan members — $0 to $38K/month. The AI systematically works the ~6,000 dormant past customers the company already owned, turning a dead asset into ~$61K/month of near-free revenue. Financing prequalification and on-site upsell context lifted average ticket from $480 to $710.
Owners don't trust marketing they can't see. One dashboard runs true closed-loop attribution — spend → call → booked → showed → completed → revenue → lifetime value — surfacing revenue-per-truck-hour, blended cost-per-lead, and membership ARR live, plus a running feed of every surge caught, call recovered, and customer reactivated. Multi-location ready.
Before turning the demand dial up, we stopped the bleed. Any call that isn't answered in a few rings gets an automatic text in seconds, a 24/7 AI receptionist books after-hours calls onto the dispatch board, and a photo-to-estimate intake removes friction. Missed-call text-back alone recovered ~$40K a month in jobs the company was already paying to generate and then throwing away.
Most marketing agencies never look here. Acquiring a customer is expensive — the profit is in maximizing each one. Automated sequences convert one-time jobs into recurring members, and the AI systematically works the ~6,000 dormant customersthe company already had on file. The reframe that sold it: you don't have a lead problem — you have a follow-up problem and a lifetime-value problem, and one system fixes both.
Seal the leaks, then turn on the demand, then squeeze lifetime value — in that order. Each subsystem is good alone; the closed loop, where completed-job data tunes demand buying, is what compounds.
Three compounding factors: capacity- and margin-aware demand buying purchases cheaper, higher-intent demand; the review flywheel pulls in free organic calls that lower blended cost-per-lead; and Never-Miss Capture converts calls that used to go to voicemail — more booked jobs from the same spend.
Almost entirely speed and persistence: 22-second speed-to-lead, instant missed-call text-back, and a 24/7 AI booker that never lets an after-hours lead go cold. The homeowner whose AC just died books with whoever answers first — and now that's always this company.
The profit is in lifetime value, not just acquisition: memberships added a recurring floor that smooths seasonality, the reactivated database produced ~$61K/month of near-free revenue, and financing plus on-site upsell context lifted average ticket from $480 to $710 — all without proportionally more trucks or staff.
Competitors bolt six tools together and optimize for cheap clicks. This is one integrated system that buys demand it can service, seals every leak, and turns each completed job into reviews, members, and reactivation revenue — which makes the next dollar of spend cheaper. Each subsystem is good alone; the closed loop is what makes it defensible.
“We thought we needed more leads. Turns out we were losing a third of the calls we got, and we had six thousand old customers we'd never followed up with. They sealed the leaks first — recovered forty grand a month I didn't even know I was losing — then turned the dial up. We went from praying for summer to revenue coming in even in the slow months. I don't think of them as a marketing company. They built us a machine.”
Owner · residential HVAC & plumbing
If calls go to voicemail, leads sit for hours, or your past customers never hear from you again — this is the system that seals the leaks and turns the dial up. Let's scope it.