Operations • For plumbing, HVAC and electrical companies

The Replacement Quotes Sitting in Your Software

Most open HVAC replacement proposals get two follow-up calls and then nothing. How AI can keep every proposal moving with notes tied to its own details, flag the warm ones, and stop chasing the homeowners who already said no.

By Justin Beegel · · 6 min read

The comfort advisor sat at the kitchen table for an hour and a half. He looked at the 17-year-old system, measured the return, checked the ductwork, and walked the homeowners through three options on a tablet. Good, better, best. Financing on the middle one. A note about a utility rebate they’d asked about.

The homeowner said they’d like to talk it over. Fair enough. The advisor called two days later and left a message. He called again the next week and got a polite “still thinking.” Then he was in four other houses a day, and the third call never happened.

Six weeks later, the first real heat wave hit, the old unit quit, and the homeowner bought from whoever answered the phone that afternoon. Your proposal is still sitting in the software, marked open.

I’ll be honest about my vantage point. I’ve never sat at a kitchen table selling a heat pump. I ran a digital agency, Infographic World, for almost 18 years, and now I use AI to fix real problems inside businesses. What I understand is how a piece of work falls between two people, and an open proposal is a textbook case: the advisor has moved on and the office doesn’t know the details.

Why follow-up stops

It’s rarely laziness. The way the job is set up makes follow-up the first thing to go.

  • The advisor is in the next house. Today’s appointments are in front of them. Last month’s proposals don’t ring.
  • The proposal sits in the software with no next step. It’s marked open, but nothing tells anyone it’s day 9 and nobody has called.
  • The second call feels pushy. If the advisor has nothing new to say, “just checking in” feels like pressure, so they stop.
  • Nobody marks the no’s. Half the open list is dead, so nobody trusts the list, so nobody works it.
  • The office can’t pick it up. A CSR doesn’t know which option the homeowner liked, what worried them, or what rebate they asked about.

Meanwhile, homeowners stall for ordinary reasons. The price was bigger than they expected. They want to see how financing works. One spouse wasn’t home for the visit. They’re waiting on a second quote. The old unit is still running, so there’s no hurry, until there is.

Each of those reasons has a specific, helpful answer. Generic follow-up doesn’t give it.

What AI-assisted follow-up looks like

It reads each proposal

The AI reads every open replacement proposal in your field service software: the three options and prices, the financing offered, the rebates or incentives the homeowner asked about, the age and condition of the current equipment, and the advisor’s notes on what the homeowner cared about. Quiet operation, a room that never cools, a budget ceiling, a spouse who wants to compare.

It builds a sequence for that proposal

Instead of one generic reminder, each proposal gets a short sequence tied to its own details. Made-up example for one household:

  • Day 2: a plain recap of the three options, with the one they leaned toward first.
  • Day 6: an answer to the question they asked, like what the monthly payment looks like on the middle option, in the exact terms your financing partner quoted.
  • Day 14: a note on the rebate they asked about, with the current terms your office confirmed.
  • First forecast heat wave: a short check-in for a system that’s 17 years old.

It drafts, and the advisor sends

Every message is drafted in the advisor’s voice, from the advisor’s name. The advisor reviews it on their phone and approves, edits or skips it in a few taps. Once you trust the drafts, the office can approve routine ones and leave the advisor the replies.

It flags the warm ones

Some proposals heat up. The homeowner opened the proposal again, replied with a question, clicked the financing link, or called about a repair on the same unit. Those get flagged for the advisor to call that day, with the summary in front of them.

It stops when it should

A clear no ends the sequence. “Went with another company” ends it and records the reason. “Not this year” moves the proposal to a seasonal check-in months out. And every proposal has a cap on touches, so nobody gets chased forever.

On tone: short, specific and useful beats clever. No fake deadlines, no “prices go up Friday” unless prices really go up Friday. The best follow-up reads like a helpful advisor who remembered the conversation.

Run the numbers

The formula in words: count the open replacement proposals from the last 90 days, multiply by your average replacement ticket, then multiply by a modest share you could recover with steady, specific follow-up.

Say, for the sake of the math, a shop has 80 open replacement proposals from the last 90 days. Say the average sold replacement is $12,000. And say steady follow-up recovers 5 percent of them. All three numbers are made up.

80 × $12,000 × 0.05 = $48,000 a quarter, from four extra jobs. Over a year that’s close to $200,000 in revenue from proposals already written, on homes already visited. Multiply by your own gross margin to see the profit side.

The recovered share is the number to be careful with. Pick one you’d defend to your banker, not one you’d put on a billboard.

To run it on your own business, pull three numbers:

  1. Open replacement proposals from the last 90 days. Not sold, not marked lost.
  2. Average sold replacement ticket. From your own closed jobs, not the price sheet.
  3. How many follow-up touches each open proposal actually got. If most got one or two, the recovered share has room to move.

Picture a Monday, before and after

This is a made-up scene. Picture the sales manager at a 20-truck HVAC company on a Monday in March.

Before. The weekly sales meeting opens with the open-proposals report. It’s 112 lines long. Nobody knows which ones are real. Each advisor scrolls their own and says “I’ll reach out to a few.” Two do. The report runs again next Monday, two lines shorter.

After. The manager opens a list of 61 live proposals, because the system and the advisors already cleared out the ones marked lost. Nine are flagged warm: three opened the proposal again over the weekend, two replied with questions, four have units over 15 years old and a forecast in the 90s next week. Each advisor’s drafts are waiting for approval. The meeting takes 20 minutes, and every proposal has a next step with a date on it.

What has to be in place first

  • Proposals in the software, not on paper. Options, prices and the selected or preferred option, attached to the customer record.
  • Advisor notes worth reading. A few lines on what the homeowner cared about. This is the raw material for every good follow-up.
  • Current financing and rebate wording. Terms change. Someone in the office owns the approved wording and updates it.
  • Contact preferences and consent. Text, email or call, and permission to follow up.
  • Outcome codes. Sold, lost to price, lost to a competitor, repaired instead, not this year. Without them, you can’t stop chasing the right people.

Guardrails worth keeping

  • The advisor approves what goes out in their name. At least until the drafts earn trust.
  • Financing and rebates are stated as the lender or program states them. The system doesn’t promise approval, rates or eligibility. Your office confirms them.
  • No invented urgency. Real weather, real price changes, real equipment age. Nothing made up to rush a decision.
  • No means no. A clear no ends the sequence, and every proposal has a cap on touches.
  • Follow the contact rules that apply to you. Texting and calling rules vary, and opt-outs get honored right away.

How you’ll know it’s working

  • Share of open proposals with a follow-up in the last seven days.
  • Close rate on proposals older than 14 days.
  • Median days from proposal to decision, sold or lost.
  • Share of lost proposals with a recorded reason.
  • Revenue from proposals that sold more than 30 days after the visit.

Where I’d start

Pull the last 90 days of open replacement proposals. Sit down with your advisors for an hour and mark the dead ones. Then have AI draft one personal check-in for each live proposal, based on its own notes, and let each advisor approve and send their own.

That single pass tells you how many were still in play. It also shows you whether your notes are good enough to build sequences on.

If this sounds like your week, send me a note through the contact page.

Questions owners ask

Won’t more follow-up annoy homeowners?

Generic follow-up does. A short note that answers the question they actually asked, like the monthly payment on the middle option or whether a rebate applies, usually reads as service. Cap the number of touches, and stop the moment someone says no.

Should the comfort advisor or the office send the follow-ups?

The homeowner met the advisor, so messages should come from the advisor’s name. The AI drafts them and the advisor approves them in a few taps. The office can own the list and the timing so nothing depends on the advisor remembering.

How long should a replacement proposal stay open?

Long enough to cover the next change in season, since a unit that limped through spring often fails in the first heat wave. After a few months with no reply, move it to a seasonal check-in rather than closing it or chasing it weekly.

Keep reading

AI at Work · · 6 min read

Your Best Tech Hates Paperwork, and It Shows in Your Cash Flow

Paperwork is where good field work turns into late invoices and lost recommendations. A tech talks into the phone after each job, AI drafts the rest, and the office reviews before anything goes out.

AI at Work · · 6 min read

Every Callback Has a Lesson Your Senior Tech Already Knew

The fix for half your callbacks is already in somebody’s head. A private assistant built on your own job history lets a new tech ask from the truck, while code and safety calls stay with licensed people.

Operations · · 6 min read

Your Dispatcher Is Solving a Math Problem by Hand Every Morning

Every morning a dispatcher matches jobs to techs by skill, location, parts and promised windows, then redoes it by 10 AM. Where AI helps, what it can’t know, and how to work out what drive time is costing you.

Want this thinking applied to your business?

Tell me what is slowing the business down, and I will tell you where I would start.