There's a rooftop unit on some strip center off the Gulf Freeway that you've been out to four times in the last fourteen months. A capacitor here, a contactor there, a pound and a half of refrigerant last spring. Each visit closed clean, invoice paid, everybody moved on. Nobody stepped back and added it up.
So the unit does the math for you. It quits at 2 in the afternoon on the first 98-degree Friday in July, the tenant's on the phone, and now you're pricing an emergency changeout with no lead time, no crane scheduled, and a customer who's sweating and shopping your number against two other companies who can "come today."
Here's the thing that stings: every signal you needed was already sitting in your own closed tickets. The repeat visits, the refrigerant top-offs, the install date pushing past fifteen years. That knowledge just never left the ticket it was buried in. This is about digging it out on purpose, before the weather does it for you.
Your service history already flagged it
A dying commercial unit almost never goes quiet. It gets loud in your records first, one ticket at a time, in a way nobody's watching because each ticket looks small on its own.
The pattern you're hunting for is boring on purpose. A unit that needed two or three real repairs in twelve months. A system you've topped off with refrigerant more than once, because a leak you keep chasing is a compressor or a coil quietly telling you it's tired. An install or manufacture date that puts the equipment past fifteen years, where the next failure is as likely to be "replace it" as "fix it."
None of that is a mystery. It's just spread across separate visits and separate months, so it never shows up as one story. Your job is to stitch the visits back into the story.
Build a watch list you can actually keep
Start dead simple. A watch list is one row per commercial unit you want to keep an eye on, and it needs maybe six columns. Fancy comes later, if ever.
- Customer and site, so you know whose roof it's on.
- Unit ID, model, and serial, so you're not guessing which of the four rooftop units you mean.
- Age, pulled from the serial or the install date on file.
- Repairs in the last 12 months, a count and a quick note on what they were.
- Refrigerant added, dates and amounts.
- Status, one word: watch, quote, or proposed.
A spreadsheet is fine to start. If your field software lets you tag equipment or run a report on repair history by unit, better, because then you're not retyping what the system already knows. The point isn't the tool. The point is that the list exists somewhere you'll actually look at it, instead of living in your head where it evaporates by Thursday.
Getting the list populated the first time is the real work, and it's a one-afternoon job if you attack it right. Pull your commercial tickets from the last year or two and sort them by site and unit. The units that show up three or four times sort themselves right to the top.
What earns a unit a spot on the list
You don't want everything on here, or the list turns into noise and you stop reading it. Keep the bar concrete. A unit earns a row if it hits any one of these, and it moves to the front if it hits two.
- Two or more repairs in the last twelve months. Once is bad luck. Twice is a trend, and a trend on a rooftop unit in Houston is a countdown.
- Refrigerant added more than once. A recurring leak you can't pin down is often the coil or the compressor. That's replacement money, not repair money, and the customer should hear it from you first.
- Fifteen years or older. Past that mark, parts get scarce, efficiency's gone soft, and one more failure tips the whole thing into "not worth fixing."
- A repair you quoted that the customer declined. That's a unit running on borrowed time with a paper trail proving they knew.
Write the rule down and stick to it. When the criteria are plain, anyone at your desk can run the list, and it doesn't depend on the one tech who happens to remember that rooftop unit's whole sad history.
Call before July does
A watch list nobody acts on is just a nicer way to be surprised. The move is to reach out while it's still spring and the roof isn't an emergency, with a planned-replacement proposal instead of a panic quote.
The pitch writes itself, because you've got the receipts. "We've been out to Unit 3 three times since last March, it's running on a system that's sixteen years old, and we'd rather plan the replacement with you than get the 2 a.m. call in August." That's not a hard sell. That's a dispatcher telling a customer the truth before it costs them.
Do it on your terms. A scheduled changeout means you order equipment with lead time, book the crane once instead of scrambling, quote real margin instead of shaving it to win an emergency, and pick a week that isn't already buried. The customer gets to budget for it instead of eating a surprise capital expense in their worst month. Everybody comes out ahead of the version where the compressor decides the timeline.
Fold it into your season. Late winter and spring, before the heat lands, work the top of the list first. A short call, then a written proposal with the repair history attached so the numbers speak for themselves. Some will say not yet, and that's fine. Now it's on their radar and on your calendar, and when it does quit, you're the company that warned them.
Where software helps, and where it doesn't
The signals we're talking about are the kind of thing a tool is genuinely good at surfacing. Counting repairs per unit, flagging anything over fifteen years, spotting a second refrigerant charge on the same system, all of that is pattern-matching across your own records, and you can set it to run on its own and hand you the list.
What it can't do is decide. A unit with three small repairs might be fine for two more summers, and a clean two-year-old unit can still throw a bad compressor. The software points a light at the units worth a look. A tech who knows that roof still makes the call on whether it's a watch, a quote, or leave it alone. Don't let a flag replace a walk on the roof.
And it won't make the phone call for you. It can tee up the list and even draft the outreach, but the conversation that turns a flagged unit into a scheduled project is yours. The tool buys you the time and the timing. You still have to use them.
Worth doing this week
You don't need a system to start. You need an afternoon and a sort function.
- Pull your commercial tickets from the last twelve to twenty-four months and sort them by site and unit. The repeat offenders rise to the top on their own.
- Start the watch list with the six columns above, even if it's just a spreadsheet for now. Add every unit that hits two-plus repairs, a second refrigerant charge, or fifteen years.
- Pick the three worst units on that list and draft a planned-replacement proposal for each, with the repair history attached.
- Make three calls before the heat lands. Not a sales push, just the honest heads-up that this one's on borrowed time.
- Set a reminder to re-run the list every quarter, so it stays a living tool instead of a one-time cleanup.
If you'd rather have this run quietly in the background and land on your desk as a ready-to-work list each month, that's the kind of thing we set up for Houston shops at AI Agency Houston. Either way, the units are already telling you which ones are next. Might as well listen before July makes you.