The company in this note is a composite. The details belong to several owners at once, so call it a mechanical contractor with about thirty-five people: three in the office, two estimators, a dispatcher, a service manager, and crews in the field.
They called about the phone. The office closed at four thirty. The phone did not. Customers find the furnace that will not light after they get home, hours after the last truck is back. Those calls went to voicemail. The owner had a feeling that work was slipping away between the call and the callback, and wanted to know if the feeling was right.
It was, a little. It was not the story.
What the interviews turned up
An AI Opportunity Assessment is two weeks inside the company, and the first week is interviews. Thirty to forty-five minutes each, every level: the owner, the service manager, both estimators, the dispatcher, the office, and four technicians. We follow real work from first contact to paid.
The after-hours phone came up in two interviews. Dispatch came up in nine.
Here is how a service call actually moved. The office took the call and wrote it on a paper ticket. The dispatcher typed the ticket into the scheduling tool. The technician got a one-line text, then called the office to ask what the customer had actually said. After the job, the technician wrote parts and hours on another paper ticket. The office typed that into accounting. Billing called the technician about two lines it could not read. The invoice went out nine days after the work.
Each of those steps was somebody's job. None of the handoffs between them was.
Estimating had the same shape, larger. A request for a quote reached the owner by email, text, or a note on his desk. He forwarded it to an estimator, who built the number in a spreadsheet, retyped the customer's details into the proposal template, then waited for the owner to look it over. The owner looked things over at nine at night. Quotes went out in five to eight days. A couple of customers a month had hired someone else by then.
Nobody was doing anything wrong. Everyone was working full days and then some. This is what happens past about twenty people: work starts moving between hands, and nobody owns the gap.
The shape of the opportunity map
By the second week every step we watched carried a tag, time sink or error risk, and a number: hours per week, by role. Then every finding went on one page, placed by impact and effort. This one had eleven items, in a shape we see often.
High impact, low effort. Do first.
- Service tickets typed once. The office enters the call one time; dispatch, the technician's phone, and billing all read the same record.
- Technicians close the job from the truck: parts, hours, a photo, a signature. Billing stops chasing them.
- Quote requests land in one queue with the customer's details already filled in, whichever way they arrived.
High impact, higher effort. The big swings.
- A first-draft estimate built from the last thirty similar jobs, for the estimator to correct instead of starting from blank.
- Proposals assembled from the estimate, so nothing is keyed twice.
Low impact, low effort. Nice, later.
- The after-hours text-back. A missed call after close gets a text from the shop's number asking what is wrong, where, and how soon. The lead is on the dispatcher's list in the morning.
Leave alone.
- Two things the owner had read about. The honest answer was a better spreadsheet, and the plan says so.
The text-back is on the map, a modest quick win: an afternoon of setup, a few hours a week back. Not the top half. The dispatch and estimating handoffs, by the math, were worth roughly ten times as many hours, and those hours belonged to the two most expensive people in the building.
The roadmap
First ninety days: the three items in the top-left corner. Tickets typed once, close-out from the truck, one quote queue. Then the text-back, because it is cheap and the owner wanted it.
The next ninety: the estimating draft, once the quote queue holds three months of clean history to learn from.
Later: proposal assembly, and whether the scheduling tool is still the right one.
Nothing all at once. Each step was sized to what the office could absorb without hiring, and reviewed with the people who would live with it before it went in the plan. The owner argued with two of the numbers. One went down.
The phone was the symptom. The handoffs were the problem, and you cannot see them from the owner's chair. You have to sit with the dispatcher.
What this is not
Nothing in the plan required buying software from us. Hand it to your own people or to us; it is the same list in the same order. We charge for the thinking. Building is optional.
If your company has a phone that rings after everyone leaves, it is worth a look. If it also has a dispatcher, two estimators, and an owner quoting at nine at night, it is worth a bigger one. The first call is free, and forty-five minutes is enough to tell which.