What Happens to a Roofing Company's Missed Calls, and How They Become Booked Inspections

Every lead that did not book goes into exactly one lane by how it arrived: called and did not book, Google Local Services, the website form, texted first, or a Meta ad. Each lane keeps its own timing. The rule across all five is the same: a next step, a named stop, or a pause.

The fix starts with sorting. Every lead that did not book goes into exactly one lane based on how it arrived, because a homeowner who filled out a form expects something different from one who called while the office was closed. Each lane keeps the timing the business already used.

Then one rule applies to all of them. At any moment a lead has a planned next text or email, a named reason to stop (booked, replied, asked us to stop, no permission to text), or a pause a person chose and wrote down. A lead with none of the three is a gap, and the system lists the gaps for the office each morning.

The messaging mostly worked. Callers were stuck on permission, because a phone call is not consent to be texted. A few replies sat for more than a week with nobody acting, and one ladder of follow-ups was not carrying leads forward at all.

I expected broken automation. Mostly I found permission and attention. The follow-up never texts someone who only called, and it should not, so callers sat with a stop reason of no text consent and no other path. A few leads replied and nobody acted for more than a week. Several planned follow-ups were sitting in held or draft past their date. And one lane's ladder of messages was not carrying leads forward at all.

Speed was not the problem people assume. Text-first inquiries and Meta leads got an answer within seconds. The first texts that looked slow had mostly landed outside daytime hours, when waiting until morning is the polite thing to do.

Federal rules require the phone owner's consent before an autodialed text reaches a mobile number, and commercial texts need written consent. So text permission gets its own box on the form, and what to do for callers who never gave it is the owner's decision.

The FCC's consumer guide is plain about it: autodialed texts to a mobile phone need the owner's prior consent, and commercial texts need written consent. A homeowner calling about a leak has not agreed to a text campaign.

So the website form asks for text permission in its own box, separate from the request, unchecked and optional. What to do for callers who never gave it is a business decision, and at A&J it is still open with the owner. I would rather leave that door shut than pretend software can answer a consent question.

The follow-up read calls, texts and its own sends, and never the email inbox. A homeowner who answered by email was invisible, so the ladder kept talking over them. The fix, built and tested, counts every reply by any channel.

This one stung. The follow-up read calls, texts and its own sends to decide whether someone had answered. It never read the email inbox. A homeowner who replied to the booking email was invisible, and every ladder kept talking over them. That is exactly how a helpful system becomes an annoying one.

The fix counts a reply on any channel. While building it I found the booking confirmation emails had been failing because two labels on each message shared the same name, and the email service rejected them. The texts had gone out, so nobody noticed. Both fixes are built and tested and ship with the follow-up lane.

Every sentence a customer will read. New wording is held until it is approved, each switch stays off until a preview run, and messages and calls still go through the existing sender and dial door.

Every sentence a customer will read. The new lane messages are written and held until they are approved. Each switch ships off and stays off until one preview run on real records. And every message still goes through the existing sender, and every call through the existing dial door, which refuses do-not-contact numbers and quiet hours. The lanes plan the next step; they do not get a second way to reach anyone.

The lanes are built, tested on the real lead list without sending anything, and live on the dashboard with their switches off. In a local run against the real records, one complete journey worked end to end on a test phone, with nothing able to reach a real customer. I am not giving you a booking-rate number, because the honest before-number does not exist yet, and a percentage built on a guess would be worse than none.

If you book almost everyone on the first call, fix nothing here. If nobody will read the morning list, the lanes only make a longer list. And if your phones go to voicemail during the day, start with how calls become booked jobs before worrying about follow-up.