How Many Jobs Before Paper Breaks? Count the Open Ones
Paper breaks somewhere around ten to twelve OPEN jobs — not ten jobs a month, ten running at once. Below that you can hold the whole company in your head, and a legal pad is faster than any software. Above it the follow-up starts dying quietly and you won't notice for a season. The fix is a system. A system is not the same thing as software, and buying the second without building the first is how contractors end up paying monthly to watch their own confusion in color.
Five a month is fine
Nobody needs to be sold on this. At five jobs a month you know every address. You know which one is waiting on the adjuster and which one the homeowner keeps pushing to next month. You know it because there are five and you are on all five. A legal pad works. A whiteboard works. Your own memory works, and it is faster than any software on earth because there is no typing and no loading and no login. Anyone telling you that you need a CRM at five jobs a month is selling you a CRM. The paper is not the problem. At that volume the paper is correct, and switching costs you speed you cannot spare.
The question that tells you where you are
Volume is the wrong measure. Jobs per month is a revenue number and it says nothing about load, because a job that closed in nine days and a job that has sat on an adjuster for eleven weeks both count as one. Count the open ones instead. Right now, today. Started and not finished and still needing something from somebody. Then answer these without opening anything.
- Name every open job from memory. Address, stage, what it's waiting on. If you can't get through the list, that is the signal — not the revenue.
- When a homeowner calls, how long until you know who they are and what you owe them? Under thirty seconds, or is it let me pull that up and call you back?
- Where does a measurement live between the roof and the estimate? Name the actual physical place. A card, a text you sent yourself, a photo of a legal pad on a tailgate.
- Who is watching the jobs that are waiting? Supplement pending, adjuster scheduled, material backordered. Nobody is working those. Somebody still has to be watching them.
- If your best rep quit this morning, could anyone else finish their jobs? Not eventually. This week.
- What did you promise a homeowner in a driveway three weeks ago that nobody wrote down?
Where the minute goes
A minute is nothing. That is the entire trap. Every piece of friction in a paper process costs about a minute, and a minute never feels like a cost while you are standing there spending it. Write the measurement on a card, type it into the estimate that night. Two minutes. Photograph the damage, then find those photos again three days later inside a roll of four hundred, then attach the right eight for the adjuster. Six minutes. Re-enter the customer into the invoicing software because it does not talk to anything else you own. Three minutes. Text a rep to ask what stage a job is at, wait, get back an answer that is partly a guess. Four minutes. None of that is a crisis. Now do about thirty minutes of it per job.
- Assumes 30 minutes of duplicate-entry friction per job and a 40-hour week. Time your own — most owners guess low, and the number is easy to collect.
- At five a month it is three quarters of a week a year. Invisible. Absorbable. You will never once notice it.
- At thirty a month it is over a month of somebody's year spent typing facts that were already written down once, by hand, correctly, an hour earlier.
- The process did not get worse. It got multiplied. That is the part nobody warns you about, because back at five a month there was nothing to warn you about.
| Jobs per month | Admin friction per year | Working weeks lost |
|---|---|---|
| 5 | 30 hours | 0.75 |
| 10 | 60 hours | 1.5 |
| 20 | 120 hours | 3 |
| 30 | 180 hours | 4.5 |
| 50 | 300 hours | 7.5 |
The roof is not an office
Every tool that fails in this trade fails the same way. It was designed by someone who has never tried to use it at one in the afternoon in August. It is a hundred and twenty degrees on that deck. You have gloves on and the screen will not register. There is glare and you cannot read anything smaller than a headline. You have one hand, because the other one is what is keeping you on the roof. Whatever that app is asking you to do up there, you are not going to do it. You are going to tell yourself you will do it later, and later you are in the truck with the air conditioning on and you have already forgotten which elevation had the soft decking. That is not a discipline problem and it will not be fixed by telling your crew to be more disciplined. It is a design problem, and it is the whole test. Any system that needs a paragraph typed on a roof is not a system. It is a wish. Ask of any tool only this: how many seconds does it cost a man who is hot and holding on with one hand. Paper genuinely wins some of these. A card in a pocket beats a slow app every time, and that is worth admitting out loud, because it tells you exactly what a tool has to beat before it earns the switch.
What breaks first is not the paperwork
You expect it to break loudly. A lost file, an angry homeowner, a crew sent to the wrong address. Those do happen and they are survivable, because you can see them and you can fix them the same day. What actually goes first is the follow-up, and it goes quietly. The supplement nobody resubmitted. The estimate that was verbal in a driveway and never sent. The homeowner who said call me in the spring, and it is spring. The adjuster who has not been chased in nineteen days. None of those generate a phone call. Nobody complains about the job you forgot to work — they just hire the roofer who called them back. It leaves your pipeline without a sound. Then you book eighteen for the month instead of twenty-four and you tell yourself the market softened. You cannot audit what you never recorded. That is the real cost of paper at volume and it never appears on a single line of your P&L.
Software will not fix your company
This is the part software companies leave out, this one included. A system laid on top of a broken process gives you a broken process that syncs to the cloud. If your estimates are inconsistent, software makes them inconsistent faster and in more places at once. If nobody knows who owns a job after the sale, a pipeline board just shows you the same confusion in color. Three things have to be true before any tool is worth a dollar.
- A methodology. One way the work gets done — how a job is scoped, what a supplement requires, who calls the homeowner and when. Written down or not, it has to be the same every time. Software encodes a process. It has never once invented one.
- The work has to be good. No system rescues a bad roof. Every dollar you save on the install comes back out in callbacks, in reviews, and in a referral rate that quietly stops without telling you why.
- A team that can carry it. This is the one that actually decides whether a company scales and the one nobody wants to hear, because hiring is harder and slower than buying. Tools multiply a team. Zero times anything is still zero.
What to do before you buy anything
Every operator who has made this jump well describes it the same way: the tech removed the minutes so the people had room to do the work. That is all it does. It is worth a great deal, and it is worth nothing at all if the three things above are missing. So do this first, in this order, and notice that none of it costs money.
- Count your open jobs today and write the number on something you will still have in six months. That is your baseline.
- For one week, tally every time you write the same fact twice — address, measurement, phone number, stage. Marks on a card, nothing fancy. Most owners are genuinely surprised by the total and it takes no effort to collect.
- Find the oldest open job you have and ask why it is still open. If the honest answer is that you forgot about it, you already have your answer about paper.
- Write your methodology down before you shop. One page. How a job moves from knock to paid. If you cannot write it in a page, no tool can encode it, and you will pay monthly for a system that reflects the confusion back at you.
- Only then look at tools, and judge them on how few seconds they cost your crew rather than how many features they list. What you are actually buying back is time in the sun with gloves on. Curbsight sells one of these, and the same test applies to it.
Frequently asked questions
When should a roofing company get software?
Do I need a CRM if I only do a few jobs a month?
What breaks first when a contractor outgrows paper?
How much time does paperwork actually cost a roofing company?
Sol Zendejas-Smith — Sol Zendejas-Smith worked home-services sales and roofing in Oklahoma before founding Curbsight in 2026. He built the platform he wished he'd had at the door.