HVAC · Call Tracking + AI Phone Receptionist

The Challenge

A huge share of HVAC demand happens after hours and on weekends, exactly when the office is closed. The client had no way to answer those calls, no idea how many they were missing, and no attribution on which marketing actually made the phone ring. The losses were real, but they were completely invisible.

We looked at how calls were handled outside business hours, and the pattern was costing the business every night:

• Calls outside business hours went straight to voicemail
• Customers with an emergency rarely left a message, they just called the next company
• There was no call tracking, so leadership could not see the scale of the loss
• No one could tell which campaigns actually produced calls
• Missed calls meant missed revenue, night after night, invisibly

After-hours coverage with defined limits

The whole problem was that after-hours calls hit voicemail and vanished. So we trained an AI receptionist on the company’s services and FAQs to answer instantly, any hour, and actually book the job instead of taking a message.

• Extended coverage: the receptionist was configured for evenings and weekends, with measured answer rate improving from 0% to about 95% during the reporting period.
• Qualifies the issue: the receptionist asks the right questions to understand the problem before anything is booked or dispatched.
• Books the appointment: qualified jobs go straight onto the calendar through the CRM sync, so the booking is done, not just promised.
• Flags real emergencies: a no-heat call in winter is escalated for immediate dispatch, so genuine emergencies reach a human fast.

Finally seeing which marketing makes the phone ring

Answering the calls was half the fix. The other half was knowing where they came from. We layered CallRail with dynamic number insertion over the whole setup so the client could measure which channels actually drove booked jobs.

• Source tracking: dynamic number insertion tags every website call, and unique numbers track offline channels, so each call carries its source.
• Stronger attribution: dynamic number insertion and campaign numbers added source data to covered calls, while repeat, saved-number, and offline behaviour remained possible limitations.
• Call recording: recordings feed the weekly quality review, so the AI flow keeps getting sharper over time.
• Spend that follows results: with real attribution in hand, the client could shift budget toward the channels that actually drove booked jobs.