It's Thursday afternoon and someone at your office is playing detective. There's a small stack of tickets from Monday and Tuesday that never got closed out clean, and now the invoices are due. What parts went on that rooftop unit? Was it two hours of labor or three? Did the tech add refrigerant, and how much of it?
So the office texts a tech who's already on another job and only half remembers. The invoice finally goes out Friday instead of Tuesday, and it goes out light, because nobody wrote down the extra hour or the second capacitor. Multiply that by a full crew across a busy week in July, and you're leaving real money on the counter and waiting days longer to bill for it.
The fix isn't a lecture about paperwork. It's making the close-out so fast that a tired tech will actually do it from the cab before pulling off the site.
Decide what a closed ticket has to include
Right now "closed" probably means the tech marked the job done and drove off. That's the problem. A closed ticket should mean the office has everything it needs to bill in full without a phone call.
Write down the short list of things that have to be captured before a ticket can go to "closed." Keep it to what actually changes the invoice:
- Parts used, with quantity. A capacitor, two pounds of refrigerant, a contactor. If it left the truck, it goes on the ticket.
- Labor time on site, and whether any of it was after-hours or overtime.
- One photo of the finished work, plus the nameplate if you don't already have the model and serial.
- Any follow-up, like a part on order or a recommended repair the customer passed on.
That's it. Four things. If the tech can't answer those four, the ticket isn't done, and the system shouldn't let it close.
Keep the whole thing under two minutes
Here's the honest truth about crews and forms: if the close-out takes longer than the drive to the next stop, it won't happen. It'll get skipped, saved for later, and reconstructed from memory. Speed is the whole game.
So build the close-out to be brutally short. A single text thread, or a form with maybe five fields, that the tech fills out sitting in the cab with the AC running before they roll. Pre-fill everything you already know: the customer, the address, the tech's name, the truck's parts list so they tap instead of type.
A few things that keep it fast:
- Let them pick common parts from a list instead of spelling them out.
- Default the labor time to the gap between "arrived" and "done" and let them adjust it.
- Make the photo one tap, straight from the phone camera.
- Never ask for anything the office can look up on its own.
If a tech has to log into three apps and scroll, you've already lost. One thread, two minutes, done.
Let the tech talk instead of type
Typing on a phone at the end of a long day is where close-outs go to die. Thumbs are slow, the screen's covered in fingerprints, and a guy who's been on rooftops since 6 AM is not going to peck out a paragraph about what he did.
Talking is faster and it's natural. This is where AI transcription earns its keep. The tech hits record and just says it out loud: "Replaced the run capacitor and the contactor, added two pounds of R-410A, two and a half hours on site, unit's cooling fine, recommended they replace the fan motor next visit." The AI turns that voice note into text and, better yet, pulls the pieces into the right fields. Parts here, labor there, the follow-up note in its own box.
Thirty seconds of talking beats five minutes of typing, and you get more detail because the tech isn't rationing words to save his thumbs. That detail is what makes the invoice complete.
Turn the close-out into a drafted invoice
Once the four things are captured, the office shouldn't be building the invoice from scratch. The close-out should draft it for them. Parts and quantities map to your price list, labor maps to your rate, after-hours flips to the after-hours rate, and out comes a draft invoice waiting for a human to glance at and send.
Draft, not auto-send. That word matters. Somebody in the office still reads it, confirms the pricing looks right, and hits send. But they're editing a nearly finished invoice instead of interviewing a tech about a job from three days ago. What used to take Friday afternoon now takes a few minutes the same day the truck leaves.
Be honest about where it still needs eyes
AI transcription is good, not perfect. It'll mishear a model number, turn "R-410A" into nonsense now and then, or drop a decimal on the refrigerant. It has no idea what your parts cost. So this doesn't remove the office from the loop, it just changes their job from reconstruction to review.
Keep a person checking every drafted invoice before it goes out, at least at first. Watch for the part numbers and the quantities, since that's where the money is. After a few weeks you'll know which fields the AI nails and which ones need a second look, and you can tighten the process around that. The goal is a fast, full draft, not a hands-off machine.
Worth doing this week
You don't need new software to start. You need a habit and a short list. Here's where to begin:
- Write your four-item close-out list and tape it to the dash of every truck. Parts, labor, photo, follow-up.
- Set the rule out loud at the next crew meeting: a ticket isn't closed until those four are in, and nobody's off the clock before it's done.
- Pick one tech and one week to test a voice-note close-out, even if it's just a text thread you review by hand at first.
- Compare that tech's invoices to the crew's. Look at how fast they went out and whether they came in fuller than usual.
- If it holds up, wire the voice note into an AI transcription tool that drafts the invoice, and we're glad to help you set that up.