Most roofers have been pitched “AI” this year, and it’s nearly always the same product: a chat bubble on the website. So most roofers assume that’s what AI for a roofing company means. A chatbot nobody uses.
The money isn’t there. It’s in two windows nobody has time for during storm season: the ten minutes after a storm lead calls, and the two weeks after the estimate goes out.
Say hail comes through on a Tuesday evening. Here’s where AI earns its keep that week, and where it has no business being.
Tuesday night: the phones start
The storm passes at 7pm. By 8pm you have more calls than people. Voicemail is where storm leads go to die, because the homeowner just calls the next truck in the search results.
An AI receptionist fixes the part that matters: every call gets answered, and every caller gets asked the same questions in the same order:
- What happened (hail, wind, a tree, a leak they noticed)
- The address, so you can check it’s in your area and map the storm path
- Is water coming into the house right now?
- Whether they’ve called their insurance yet
- Best number, and is texting OK
Then it books an inspection from your real calendar and texts a confirmation.
What it should not do: quote a price, guess whether insurance will cover the roof, promise a crew date, or give roof-safety advice beyond “please stay off it.” If water is coming in now, it transfers to your on-call person live. More on the hand-off line in AI receptionists for contractors.
Wednesday morning: triage before the trucks roll
You wake up to a list. Without a system, that list is voicemails, texts to three phones, and sticky notes.
With lead intake set up, every overnight caller is already in your CRM, tagged with what they said. Water-in-the-house calls are at the top. Then visible damage. Then “my neighbor got a new roof and I’m wondering.”
AI sorts the list. A person decides who gets a truck first. You know which streets took the worst hit.
Daytime overflow runs the same way. When the office is on one line and two more ring, a missed-call text goes out within a minute so the homeowner knows you saw them. Setup is in missed-call text-back for contractors.
Thursday and Friday: estimates go out, then silence
By Friday you’ve sent a dozen estimates. Now comes the part roofers hate: the homeowner is collecting three bids, the adjuster hasn’t been out yet, and nobody calls you back.
This is the second window. Estimate follow-up drafts a check-in a few days later, written in your voice and using the details of that roof: what you found, the photos you took. Then another a few days after that. You approve each message before it sends.
The specifics are what make it work. “Just checking in” gets ignored. “Here are the two photos of the cracked shingles on the north slope” gets a reply.
No fake urgency. Don’t say the schedule is filling up unless it is. After a real storm it usually is, so say so plainly.
The next week: insurance paperwork
The claim side is where hours vanish: scope write-ups, a cover letter to the adjuster, a supplement when the adjuster’s scope misses the drip edge and the ice-and-water shield.
AI is good at the first draft of all of that. Give it your inspection notes and photos, and it produces an itemized scope summary and the supplement letter in your format. A human reads and approves every page before it goes to a carrier or a homeowner. No exceptions: this is money, and someone else’s claim.
What AI does not do: decide what’s covered, negotiate with the carrier, or tell a homeowner their claim will be approved. When a claim gets ugly, a public adjuster or an attorney is the right call, not a prompt.
Two weeks later: the final walkthrough
The homeowner is on the lawn looking at a new roof. That’s the moment to ask for a review, and the moment everyone forgets.
Set the trigger to “job marked complete.” The same day, a text goes out with your Google review link. One reminder a few days later if nothing came in.
What stays human
The rule we build around: anything that touches money or a client waits for your approval, and every automated run is logged. AI answers the phone at 9pm, sorts the list, drafts the follow-up and the supplement. You decide who gets a truck, what the roof costs, and what goes to the adjuster. More in roofing automation.
Where to go from here
Here’s a storm-season lead-response SOP you can put on the office wall this week. It works with an AI receptionist, an answering service, or a cousin with a cell phone.
Who answers:
- Business hours: the office line. Overflow gets the text below.
- After hours: the AI receptionist or on-call phone. Water coming in goes live to on-call.
- Nobody picks up: the text below goes out on its own within one minute.
How fast:
- Text within one minute of a missed call.
- Human callback within 30 minutes during business hours.
- Overnight leads get a call before 8am the next morning.
The first text (edit so it sounds like you, and only promise windows you can keep this week):
Hi, this is [Name] with [Business]. We got your call about storm damage at [address]. Is water coming into the house right now? Reply YES and we’ll call you back within the hour. If not, reply with a good time for a free inspection and we’ll get you on the schedule.
If nobody answers:
- 30 minutes with no callback: the lead moves to the second person on the list, with a ping to their phone.
- Two hours with no callback: it goes to the owner.
- Every lead lives in one list with a name next to it. A lead with no owner is a lead you paid for and lost.
Run it for one storm and count how many leads got a callback inside the window. That’s your baseline.
If you’d rather have it running on your phone line and CRM instead of on paper, book a free 30-minute call. We’ll walk through how we’d set up this exact SOP in the tools you already use, and you’ll leave with a written plan and a fixed price before you decide anything.