6 min read
This is an illustrative story. Dev is a composite based on the way small seasonal service businesses commonly work, not a real person, and the numbers are realistic examples rather than reported results.
It is January 3rd, and Dev’s phone has 41 unread texts. Every one of them says some version of the same thing: “When are you coming to take the lights down?” Dev runs a one-person holiday lighting business. He spent November hanging lights on 58 homes, and now all 58 of them want the lights gone at once, in a two-week window, before the HOA letters start arriving.
Last year that window cost him three Saturdays of driving back and forth across town, a handful of annoyed customers, and one very cold evening on a ladder in the dark. This year, he decided to fix it in November, with one free AI assistant and about two hours.
The real problem was the order, not the work
Taking lights down is quicker than putting them up. Dev’s problem was never the ladder time. It was three things around it:
- Booking by text. Every customer picked their own day, so he zigzagged across town.
- No reminders. People forgot he was coming, left cars in the driveway, or locked side gates.
- No record of what went where. Which clips were his, which extension cords belonged to the homeowner, which house had the custom roofline kit.
He did not want new software. He already had a spreadsheet of customers and a phone. So he used Microsoft Copilot, the free AI chat assistant he already had on his laptop, to do the thinking and writing, and kept everything else the same.
Step one: group the addresses into route days
Dev pasted his 58 street names and neighborhoods (no customer names or phone numbers) into Copilot and asked it to group them into clusters of six to eight homes that sit close together, using the neighborhood names as a guide. He asked for each cluster to get a label like “North Hills A”.
The first answer was not perfect. Copilot put two streets in the wrong group because they share a name with a street across town. Dev caught it because he knows his town, and he told Copilot to move them. That is the pattern that made this work: the AI drafted, he corrected. If you rely on AI for anything location-based, our guide to catching AI hallucinations explains why a quick human check is not optional.
He ended up with eight clusters. Each one became a takedown day, and within each day he used Google Maps’ multi-stop directions to put the houses in driving order.
Step two: let customers pick a day, not a time
Instead of asking “when works for you?”, Dev asked Copilot to write a short, friendly text for each cluster. The message told customers which two days their street was scheduled for and asked them to reply 1 or 2. It also said, in one line, what he needed from them: driveway clear, side gate unlocked, and any homeowner-owned cords left plugged in so he would leave them.
He asked Copilot for three versions and picked the shortest. Then he added one line in his own words, thanking them for the season. He sent the texts on December 20th, which meant January was planned before the rush began.
Result: of 58 customers, 51 replied with a number within a week. The other seven got one follow-up, also drafted by Copilot, and five of those answered. He took the last two as “whenever you are in the area”.
Step three: reminders that answer the questions before they are asked
Last year’s reminders were Dev remembering, at 7 a.m., to text people. This year he asked Copilot for one template reminder to send the evening before each takedown day. It had a blank for the street name and repeated the three things he needed (clear driveway, open gate, cords left plugged in).
He also asked for a separate message for the day after, saying the lights were down, that clips had been removed from the gutters, and that he was opening his calendar for next November with a small early-booking discount. That last line turned his takedown visit into his first sale for the following season.
If you run any appointment-based business, the same idea works for no-shows. Our story about a solo mobile dog groomer cutting no-shows shows a similar reminder system.
Step four: a two-line record per house
The last piece was memory. During installation in November, Dev had dictated quick voice notes on his phone after each house: “Kowalski, white C9s, their cords on the porch, our timer.” In December he pasted the transcribed notes into Copilot, without the customer names, and asked it to turn each note into two lines: what is his, and what stays.
He pasted those two lines into a new column in his spreadsheet. On takedown day, he read the line before he climbed the ladder. No more removing a homeowner’s extension cords and having to drive back with them.
What changed
- Takedown days: from a scattered 13 days of zigzag driving to 8 planned cluster days, plus one catch-up day.
- Texts in January: from dozens of “when are you coming?” messages to a handful of replies to his own plan.
- Return trips for mistakes: from several last year to none.
- Next-season bookings: 14 customers booked November installs during the takedown week, because the follow-up asked them at the right moment.
Total setup time in November and December: about two hours. Total cost: nothing beyond tools he already had.
What you can take from Dev’s winter
You do not need to hang lights for this to apply. Any business with a seasonal rush has the same shape: a burst of work, customers who all want the same window, and details that live only in your head. The lessons:
- Plan the rush before it starts. Dev’s real win was sending the plan on December 20th instead of reacting on January 3rd.
- Offer choices, not open questions. “Reply 1 or 2” gets answers. “When works for you?” gets negotiation.
- Keep personal details out of the AI. Dev pasted streets and notes, never names and phone numbers, and did the matching in his own spreadsheet.
- Correct the draft. The AI got two streets wrong. A quick read fixed it.
If your own January is a quieter month, it is also a good time to clean up the customer list itself. Our guide on cleaning a messy customer list with AI pairs well with this.
What is the rush your business dreads every year? Tell us in the comments, and we may build the next illustrative story around it.



