How a 3M HVAC Shop Grew 78% in One Year
A family-owned HVAC contractor moved from fragmented tools and manual workflows to a custom AI back office. In one year, jobs grew 78 percent, customers grew 52 percent, and they finally knew where their money came from. This is what we built, and why it worked.
2025 to 2026. Client name kept private at their request.
A $3M operator running on spreadsheets and hope.
Profitable, but fragmented. Three separate software platforms that did not talk to each other. Lead generation scattered across Google Local Services and Facebook ads. Dispatch done by phone calls and text. Customer follow-up was reactive, not systematic. No way to know which ads actually worked. No way to know which ads actually worked. Visibility into ROI: zero.
No central customer data
Customer information lived in email, text threads, and spreadsheets. The crew had no visibility into service history, so they pitched upgrades the customer already had. Repeat customers were treated like new leads.
Pixel was broken
The Meta Pixel was installed incorrectly. Facebook was showing 1 conversion for every 7 that actually happened. All ad decisions were made on data that was 7x wrong. Good ads got killed. Bad ads got scaled.
Leads fell through
No centralized system to capture and follow up on inbound leads. If the office manager was busy or away, leads went unanswered. Customers called back the next shop on the list instead.
Manual everything
Dispatch, follow-up texts, quote requests, appointment reminders, and upsell prompts were all manual work. The owner was the bottleneck. They could not scale without hiring more people.
Tired of SaaS rent. Ready to own.
The owner called because he had heard about Forge Built from another contractor. He was tired of monthly software bills that never stopped and never left him owning anything. The conversation went like this: 'What would it cost to build something we actually own?' He was skeptical when we quoted him $25,000 flat for a 6-week custom build. Then we did the math on renting. Mid-tier field service platforms run hundreds to over a thousand dollars a month, every month, forever. A few years of that buys a build outright. And at the end of renting you own nothing. No data. No code. No control. The Forge model made sense immediately: one-time build, complete ownership, optional $3,000 per month maintenance if he wanted our support. He signed.
Six systems wired into one machine.
The build took 6 weeks. Week 1 was discovery and scoping. Weeks 2-4 were custom build. Weeks 5-6 were testing, refinement, and training. Here is what went live.
Capture every lead in seconds
- Google Local Services Ads fed directly into the CRM
- Facebook and Instagram ads with fixed pixel data
- Lead scoring logic: hot vs. future vs. bad fit
- SMS auto-response: acknowledgment within 2 minutes
Sequences that never sleep
- Job reminder sequence: confirmation, 24-hour reminder, follow-up after completion
- Upsell engine triggered by customer behavior
- Re-engagement campaigns for dormant customers
- Customer feedback collection and review requests
Jobs assigned and routed automatically
- Live crew visibility: each tech sees their next job, customer sees who is arriving
- Dynamic routing by distance, skill match, and availability
- Photo capture and job notes auto-saved
- Service history accessible to every crew member
AI handles the first response
- AI triage on inbound calls and texts: emergency vs. scheduling vs. quote
- Ticket queue by urgency: emergencies first, others by arrival
- Escalation to human on complex issues
- Auto-response templates for common questions
Revenue triggered by behavior
- Recommendation logic: 20-year-old furnace gets maintenance plan offer
- Crew-facing prompts with customer history before arrival
- Financing options if price is an objection
- Performance tracking: which upsells close, which crew converts best
Dashboards that tell the truth
- Revenue by source: LSA vs. Facebook vs. repeat customer vs. referral
- Cost per acquisition by channel
- Pipeline health and revenue forecast
- Customer lifetime value and crew productivity
What the numbers show.
Six weeks after launch, the system was live. Twelve months later, here is what moved.
Jobs went from 159 to 283. Customers went from 168 to 256. The best-performing ad creative came in at 26 dollars per lead, and the account ran a 4.12 percent click-through rate, roughly double typical contractor rates.
No new headcount. One owned system.
The contractor did not hire more people. They did not spend more on ads. They built infrastructure that ran the parts a person kept dropping. Here is the breakdown.
No lead leaks
Every inbound call, form submission, and message got captured and answered in seconds, day or night. Customers stopped calling the next shop on the list because this one actually answered.
The pixel is fixed
Conversion tracking moved from 7x undercount to accurate. Ad dollars stopped being allocated to the wrong creative. The best ads got scaled instead of killed by mistake.
Crew has the full story
Every technician sees customer history before they arrive. No duplicate pitches. No missed upsell opportunities. Service history visible means every job is a potential retention moment.
Automation handles repetition
Follow-up texts, appointment reminders, quote follow-ups, review requests, and upsell prompts now run without anyone touching them. The owner is no longer the bottleneck.
If you are doing 2M to 6M, this is your roadmap.
You are probably losing revenue in the same places this contractor was. Fragmented tools. Bad data. Manual follow-up. The pixel might be wrong. Customer data is scattered. You have no single source of truth for where money comes from. The build we did for them is the same build we do for anyone in their revenue range. Custom to your workflow. Built in 6 weeks. Yours to own. No monthly rent. No lock-in. A system that works exactly like your business works, not one where you conform to a platform.
This case study is built on verified data from three sources: Meta Ads Manager (ad performance, spend, CTR, conversions), the field-service management platform (jobs, customers, timestamps), and the CRM (pipeline data, follow-up sequences, customer history). All numbers are from real dashboards, real exports, and real business books. No projections. No math that assumes things that did not happen. The client name is kept private by their request. Everything else is exactly as it happened.
See what we would build for your HVAC, plumbing, or electrical business.
Thirty minutes. No pitch. We map exactly where AI fits your workflow and hand you the roadmap, whether you hire us or not.
Book the audit call