A 12-person operation was losing jobs to faster competitors. They deployed AI Receptionist in a single afternoon. Here’s the full 90-day breakdown.
Piedmont HVAC ran a tight operation. Twelve employees, a solid reputation, and years of repeat customers. But as the team grew and call volume picked up, something broke: the phone.
During peak season, when every homeowner in a 40-mile radius is either sweating through a broken AC or shivering through a dead furnace, the phones were ringing nonstop. One office manager handled everything. When she was on a call, other calls went to voicemail. When she was out on a job site coordination call, same thing. When she was at lunch, same thing.
The owner, Marcus, started noticing a pattern. He'd return missed calls and get: "Oh, we already called someone else." In the HVAC business, that's not a scheduling inconvenience. That's a $300–$1,200 service call walking out the door, sometimes the start of a long-term maintenance contract worth far more.
He ran the numbers. Estimating conservatively, they were losing 8–12 calls per week during peak months to competitors who simply answered faster. At an average ticket of $450, that was $3,600–$5,400 in revenue walking away every week. The office manager was already maxed out. Hiring another front desk person would cost $2,800–$3,200/month before benefits.
“I’d call people back 20 minutes after they left a voicemail and they’d already booked with someone else. That’s when I knew we had a real problem.”
Marcus T., Owner, Piedmont HVACMarcus had been on Nextiva for his business phone system for two years. He heard about the XBert AI Receptionist add-on from another small business owner at a local chamber of commerce meeting. The pitch was simple: $99/month, answers every call, routes intelligently, works with the system you already have.
His criteria were straightforward:
He signed up on a Tuesday afternoon. By 5pm, the AI Receptionist was live.
Setup took about 3.5 hours, most of which was writing the custom greeting script and configuring the routing rules. The technical side, connecting to the existing Nextiva account, was essentially automatic.
The routing logic Marcus configured:
The greeting was customized to feel local: "Thanks for calling Piedmont HVAC, we've been keeping homes comfortable in the [region] since 2009. How can we help you today?" Not robotic. Not generic. On-brand from the first word.
Zero calls to voicemail during business hours. After-hours calls now captured with full contact info instead of blank voicemails. Office manager reported feeling "like a weight lifted", no more scrambling to return a backlog of unclear voicemails.
Two issues surfaced: callers saying "it's getting really hot in here" weren't being caught by the emergency routing (they needed to say "no AC" specifically). Marcus updated the trigger phrases. Also added a seasonal FAQ about tune-up pricing to reduce basic inquiry call transfers to the office manager.
Marcus tracked callbacks from AI-captured messages. Of 63 after-hours calls captured in month 2, 41 converted to booked appointments. Previously, those calls would have been missed entirely or reached an unclear voicemail. At average ticket $450, that's approximately $18,450 in recovered revenue in a single month.
Extended AI Receptionist to handle the company's second line (used for commercial accounts). Configured separate routing trees for residential vs. commercial callers. The office manager now handles zero routine incoming calls, only transfers from AI that genuinely require her attention.
Marcus is a straight-shooter. He shared the actual financials, not a sanitized version.
Cost of AI Receptionist: $99/month. Month one through three combined cost: $297. In that same period, he tracked $47,200 in revenue from appointments that came through calls captured by the AI, calls that would previously have gone to missed voicemail or been picked up by a competitor. He acknowledges not all of those would have been lost without the AI, but estimates 60–70% wouldn't have converted without instant answer capability.
The office manager's time reclaimed: approximately 2.5 hours per day previously spent managing inbound call chaos. She now uses that time on scheduling, customer follow-up, and the commercial account coordination work Marcus had been putting off hiring for.
The math on hiring a second front desk person: never happened. Not needed.
Real case studies include the real friction. Three things didn't go perfectly:
Accent recognition had a noticeable failure rate in the first two weeks. Callers with heavier Southern accents were occasionally being asked to repeat themselves by the AI. This didn't cause call drops but did add friction. It improved somewhat on its own by week three, the AI appears to adapt, but was never fully eliminated.
The emergency routing required more fine-tuning than expected. The initial keyword set was too narrow. Marcus spent about 90 minutes in week three expanding the trigger phrase library. Worth doing, just not instant.
Older customers, particularly in the 65+ demographic, occasionally expressed frustration with an AI greeting. Marcus's solution: configure the AI to immediately offer a "press 0 to reach our team directly" option for any caller who hesitates. That resolved most of the feedback.
For $99/month, Piedmont HVAC eliminated missed calls, recovered an estimated $47,000+ in first-quarter revenue, freed up a key employee for higher-value work, and avoided a $3,000/month hire. The ROI was not close. If your business depends on inbound calls and you're still relying on a single human to answer them, this is the most straightforward fix in the stack.
Set up the emergency routing before going live, not after. The two weeks of narrower keyword matching cost some calls that should have routed to the on-call tech immediately. Twenty minutes of upfront configuration planning would have avoided that.
Also: turn on the "press 0 for a human" option from day one. Don't wait for customer feedback to surface the need. For field services businesses with older customer demographics, that option is table stakes.
The best AI tools, real case studies, and actionable guides, delivered every Thursday. No noise. Just signal.