A home services company in the Tampa Bay area runs three divisions: plumbing, electrical, and HVAC. Fourteen technicians total. Five plumbers, four electricians, five HVAC techs. They handle about 60 service calls per day across all divisions.
The dispatch process was managed by one person — Donna — who had been with the company for eleven years. Donna knew every technician's skills, speed, location tendencies, and personality. She knew that Mike the plumber was fast with water heaters but slow with drain cleaning because he was thorough to a fault. She knew that Janet the electrician was the only one certified for commercial panel upgrades. She knew that the HVAC team's newest tech should not be sent to the older Gulf-front condos because the units in those buildings required experience with legacy systems.
Donna managed all of this on a whiteboard, a paper calendar, and her phone. When a call came in, she identified the service needed, checked the whiteboard for technician availability, considered drive time, and assigned the job. It worked because Donna was exceptional.
It failed whenever Donna was not there.
When Donna took a sick day, the owner dispatched. He assigned jobs based on whoever was next on the list. Technicians drove 45 minutes to a job that was 10 minutes from another technician's current location. An electrician showed up to a commercial panel upgrade he was not certified for and had to leave without doing the work. Three calls went to voicemail because the owner was on the phone with a technician who was lost.
The owner estimated that non-Donna dispatch days cost the company $800 to $1,200 in lost productivity, wasted drive time, and missed calls. With Donna taking vacation, sick days, and personal days, that added up to 15 to 20 non-Donna days per year — $12,000 to $24,000 annually. And that assumed Donna never left the company.
But the bigger cost was subtler. Even on Donna's days, the dispatch was not optimal. It was good. Donna's experience and judgment produced good results. But she was making 60 decisions per day based on a whiteboard she could see and a mental model she could hold. She could not simultaneously optimize all 14 technicians' routes across all 60 jobs. Nobody can.
What the tool does
The tool manages dispatch by combining the knowledge Donna had in her head with the optimization capacity that no human can match at scale.
Every technician has a profile: their certifications, their specialties, their average completion time by job type, their home base location, and any restrictions (the new HVAC tech's limitation on legacy systems, Janet's exclusive certification for commercial panels).
Every incoming job is classified: service type, estimated duration based on the customer's description and historical data for that job type, location, urgency level, and any special requirements (commercial certification, specific equipment needed).
When a new call comes in, the tool identifies which technicians are qualified for the job, where each qualified technician currently is (based on their last completed job or their current job's estimated completion time), the drive time from each technician's current or projected location to the new job, and each technician's remaining schedule for the day.
It assigns the job to the technician who produces the best outcome across three factors: shortest customer wait time, least additional drive time, and best skill match for the job type.
The dispatch coordinator — still Donna — sees the assignment and can approve or override. The override is important because Donna still has context the tool does not. She knows that the customer on Oak Street is elderly and prefers Mike because he is patient. She knows that the job on Kennedy Boulevard is at a property with a difficult parking situation that the newer techs struggle with.
What changed in the first month
Average drive time between jobs dropped from 28 minutes to 17 minutes. The technicians were spending 38 fewer minutes per day driving, which translated to approximately one additional job per technician per day.
Across 14 technicians, that was 14 additional jobs per day. At an average ticket of $185, the additional capacity was worth approximately $2,590 per day — $51,800 per month.
The company did not immediately capture all of that capacity because they did not have 14 additional jobs waiting every day. But they captured enough to increase daily completed jobs from 58 to 67 in the first month, representing $1,665 per day in additional revenue.
Missed calls dropped to near zero because the tool integrated with the phone system. When a call came in and the coordinator was on another line, the tool captured the customer's information, classified the likely service need based on the phone number (existing customer) or the IVR selection (new customer), and queued it for assignment. The coordinator confirmed the assignment when she was free instead of the customer hearing voicemail.
What Donna said
"The first week, I hated it. I felt like someone was telling me I was not good enough at my job. By the second week, I noticed I was not exhausted at 3 PM. I used to be mentally fried by mid-afternoon because I was holding 60 jobs and 14 people in my head all day. Now the tool holds most of it and I hold the parts that need a human touch."
"I override it maybe 5 times a day now. At the start, I overrode it 15 to 20 times a day. The tool learned from my overrides — or maybe I learned to trust the tool's assignments. Probably both."
The non-Donna day test
After three months, Donna took a planned vacation day. The backup coordinator, who had been trained on the tool but had none of Donna's institutional knowledge, managed dispatch for the day.
Sixty-one jobs were completed. Average drive time was 19 minutes — 2 minutes more than Donna's average with the tool but 9 minutes less than the owner's average without it. Zero missed calls. One minor assignment issue where a technician was sent to a job that required a part he did not have on his truck — a problem Donna would have anticipated but the backup coordinator did not.
The owner told me: "That was the day I stopped worrying about what happens when Donna retires. The tool does not replace Donna. But it means we are not one resignation away from chaos."
The cost
Eight days of build time. The tool integrates with the company's existing scheduling software and phone system. GPS location is pulled from the technicians' work phones (with their knowledge and consent). No per-job fees. No monthly platform subscription.
The owner's ROI calculation: the additional completed jobs in the first month alone exceeded the tool's build cost. Every subsequent month is pure margin improvement.
Get posts like this in your inbox
No spam. New articles on AI strategy, governance, and building with AI for small business.