Tomorrow's an exterior day. Roof tear-off, a full repaint, a paver patio, a truck full of mulch and cleanups. The crew's set, the customer's expecting you, and the whole day hangs on the one thing you don't control: the sky.
Then it flips overnight. You wake up at 5:40 to rain on the glass and a radar that's solid green from Worcester to the coast. Now there's a crew rolling toward the lot, a customer who has no idea, and a morning of "are we still on?" texts flying both ways while nobody's turning a wrench.
Here's the part that stings. The forecast knew. It knew around nine last night, while you were on the couch. Seeing rain coming isn't the problem. The gap is that nothing reached out to your customers in the evening and started moving the job. That's a fixable gap, and you can close it tonight.
Pick your go/no-go line before the season starts
Rain isn't a yes or no. A quarter inch at noon is a different animal from a passing shower at 6 AM that's gone before you're on the roof. And every trade has a different breaking point.
So decide the number now, when you're not staring down a crew in the lot. Write it on paper. A few examples of what a line looks like:
- Painter: anything over a 40% chance between 7 and 4, or temps that won't clear 50, because the paint won't cure and you'll be back to redo it.
- Roofer, tear-off: 50% or more, or any steady rain forecast before noon. An open roof and a surprise cell is a very bad afternoon.
- Mason: rain plus overnight temps near freezing. Fresh mortar and frost don't mix.
- Landscaper: a higher line. You can mow damp, but a soaked install day tears up the lawn and buries the crew in mud.
The point isn't these exact numbers. It's that you set yours once, per crew and per job type, so the decision is already made when the forecast rolls in. No debate at dawn.
Let something watch the forecast so you don't have to
Checking the weather for every job, every night, by hand? You won't do it in July, and you definitely won't do it during the spring rush. So hand it off to a small automation.
Here's the shape of it. Each evening, around seven or eight, a simple program looks at tomorrow's jobs, pulls the forecast for each job's town from a weather service (the same data the apps use, just read by software instead of you), and checks it against the line you set. Any job that busts the line gets flagged.
From there you've got two ways to run it:
- Review first: the system texts you a short list at 7 PM. "3 jobs tomorrow look wet: Riley, Nguyen, the Davis repaint." You tap approve, it sends the customer texts. Good for borderline forecasts where you want the last call.
- Auto-send: clear washout, high confidence, it texts the customers itself and copies you. Good for the days there's nothing to argue about.
Most shops want review-first for a while, then let the obvious washouts go automatic once they trust it. Either way, the machine does the watching. You keep the judgment.
Write the text so people actually reply
A text that just says "rain tomorrow, we'll be in touch" is worthless. It leaves the customer hanging and leaves you with the same phone calls, a day later. The message has to do three things: tell them it's off, hand them a real new date, and make saying yes take one thumb.
Something like this:
"Hi Dana, it's Mike from Coastal Painting. Rain's moving in tomorrow so we can't paint. I've got you penciled for Thursday the 14th, same crew, 8 AM. That work? Reply YES and you're locked in, or tell me a better day."
Look at what that does. It names you and them, so it doesn't read like a robot. It gives a specific slot, not "sometime next week." It asks one clear question. And it offers an out if Thursday's bad, which keeps them from ghosting you.
Keep it to a few lines. People read these in a parking lot with one hand. A wall of text with an apology paragraph and a weather explanation gets skimmed and ignored.
Confirm the rebook, don't just cancel
This is the piece most shops miss. Canceling is easy. The rebook is the whole job.
If your text just kills tomorrow and promises to "call to reschedule," that job drops into a hole. You get busy, three days pass, and your customer is sitting there with a project not done and your competitor one Google search away. A canceled job is a job at risk.
So every weather text carries the next dry slot inside it. You're not asking permission to cancel. You're moving them to a new day and asking them to confirm. The difference in how it lands is night and day: one feels like "they bailed on me," the other like "they already handled it."
And yes, confirm that new date too, the night before, same system. A rebook you never confirmed is just another 6 AM guess.
Where this won't save you
Forecasts are wrong. Some nights you'll move a job and tomorrow turns out dry, and you'll feel dumb about it. That's the trade. Set your line where a wasted reschedule costs you less than a wasted truck roll with a full crew, and you'll come out ahead over a season even with a few misses.
Borderline days are still yours. When it's a coin-flip 45% and the radar's jumpy, that's a human call, not a software one. Let the automation handle the clear washouts and the clear green lights, and keep the judgment calls on your desk.
One more honest bit: this works because exterior customers mostly get it. Everyone around here knows the forecast lies. Tell them straight that rain's coming and you're moving them to the next dry day, and most will thank you for the heads-up instead of griping.
Worth doing this week
- Write down your go/no-go line for each type of job. One number for chance of rain, plus anything else that stops you (wind, temp, frost). Ten minutes, tops.
- Draft the reschedule text in your own voice. Name, specific new slot, one yes/no question, an out. Save it where your office can grab it.
- Pick your mode: review-first, where you approve a nightly list, or auto-send for the obvious washouts. Start with review-first.
- Run it by hand for one week before you automate. Check the forecast at 8 PM, send the texts yourself, watch how customers respond. You'll learn fast what wording lands.
- Then wire up the automatic check so the forecast gets watched every night without you lifting a finger.
If you want help setting up the forecast watch and the texting so it runs on its own, that's the kind of thing we do at Mass AI Agency. Either way, the move's the same: stop finding out about the rain at 6 AM. Find out at 8 PM, with tomorrow already handled.