Skip to content
HOME·BLOG·OPERATIONS

Wasted Taps: Why a Good Team Still Hits a Ceiling

Sol Zendejas-Smith·Founder, Curbsight·Updated August 2, 2026·8 min read
The short answer

Friction in a workflow is measured in taps, not hours — the number of times the same fact gets entered by a person who already knew it. A weak team turns that friction into dropped balls, which you can see. A good team absorbs it instead, which means you cannot. So the bill never arrives as a lost job. It arrives eighteen months later as the rep who was carrying it, and by then you have misdiagnosed it as a people problem.

The math that does not add up

You hired well. You have a closer who does not need chasing, an office person who catches things before you do, and crews that show up. By every measure you have the team you said you needed two years ago. And the ceiling did not move. You are doing about what you were doing, with more people, and you cannot point at the reason. When there is no obvious villain, most owners land on the same three: the market got harder, the leads got worse, or somebody is not pulling their weight. Sometimes one of those is true. Usually the answer is duller and closer to hand.

Count taps, not hours

Hours are the wrong unit because nobody experiences their day in hours. They experience it in interruptions. So count taps instead — every discrete action a person takes to move one fact from where it already exists to where it needs to be. Not the work. The moving of the work. Take one job you closed last month and trace it from the yes at the door to the money in the account. Count every time a human being typed, photographed, copied, forwarded or re-stated something that was already known by somebody on your payroll.

  • Do not estimate this. Trace one real job, on paper, with a pen. Every owner who does this comes back with a bigger number than they guessed, and it is usually the fourth or fifth re-entry that lands — the one nobody would have listed from memory.
  • The tap is the unit because it is where attention breaks. A rep does not lose ninety seconds re-typing an address. He loses the ninety seconds, plus the thread he was holding, plus the call he was going to make next and now will not.
  • Then put each re-entry against the app that demands it. Most shops in this trade are running four — something for canvassing, something for photos, a CRM, and something for the back office — and none of the four knows what the other three were told. That is where the number below comes from.
The same factWhere it startsHow many times it gets re-entered
The addressThe rep, at the doorThe CRM, the estimate, the invoice, the material order, the crew text
The measurementsA card, a photo, a note-to-selfThe estimate, the proposal, the order, the supplement
The photosThe camera roll, 400 deepFound again, renamed, attached to the adjuster packet
The homeowner's numberThe call log, no name attachedThe CRM, the scheduler, the crew, the invoice
The stage of the jobIn somebody's headA text, a call, a guess, a board nobody updated
RE-ENTRY TAPS PER YEAR vs JOB VOLUME
The cost of one more app

Every app you bolt on adds a seam, and every seam is a place the same fact gets typed again. The lines below do not converge as you grow — they spread. That is the part that makes a good team feel like it is standing still.

00h24k100h48k200h72k300htapshours510152025303540jobs per monthCurbsight: 5,400 taps/yr at 25 jobs a monthCurbsight2 apps: 15,600 taps/yr at 25 jobs a month2 apps3 apps: 25,800 taps/yr at 25 jobs a month3 apps4 apps: 36,000 taps/yr at 25 jobs a month4 apps5 apps: 46,200 taps/yr at 25 jobs a month5 apps
At 25 jobs a month, running four apps instead of one costs 30,600 re-entry taps a year — 128 hours. At $100 an hour, that is $12,750 a year.
How this is built, so you can argue with it: one app still costs 3 re-entries a job, because one system is not no system. The table above counts 20 across a four-app stack, so each seam between apps carries about 5.7 — and re-entries grow in a straight line with app count, not exponentially. That is the smaller of the two honest readings: facts travel along a chain, not between every possible pair of apps. At 6 taps each and 15 seconds a tap, four apps at 25 jobs a month lands on 150 hours a year, which is the same figure the rest of this piece uses. The dollar number is yours, not ours — we do not know what an hour of your best person is worth, and anyone who tells you they do is selling something.

What a good team does to the number

Here is the part that makes this hard to see, and it is the whole reason a good team does not fix it. A weak team turns friction into dropped balls. The supplement goes unsent, the homeowner does not get called back, the crew shows up at the wrong address. That is visible. It is loud. You find it and you fix it. A good team does the opposite. They absorb it. Your closer stays until nine to catch up on entry. Your office person double-checks everything because she has learned the system will not. Your best rep quietly builds his own spreadsheet because the board cannot be trusted. None of that generates a complaint. Nothing lands on your desk. The friction is fully paid for — by them, out of their evenings — and it never appears anywhere you would look for it.

  • This is why the good-team version is more expensive, not less. Bad teams give you a bug report. Good teams give you silence and a slowly rising cost you have no line item for.
  • You will not read it in your numbers. You will read it in a resignation, and by then the cause is eighteen months upstream and looks nothing like software.

The arithmetic nobody runs

Put a number on it, because a feeling does not survive contact with a Monday. Take thirty minutes per job of pure re-entry — not the work, just the moving of it. That is a conservative figure and most shops that measure honestly come in higher. At twenty-five jobs a month it is twelve and a half hours a month, and a hundred and fifty hours a year. Just under four working weeks.

  • Four weeks. Ask what four weeks of your best person, aimed entirely at selling, would have produced. That is the number, and only you can fill it in — which is exactly why it is worth doing rather than reading.
  • Now the part owners skip: a share of those hours are YOURS. Owner-hours spent typing an address that was already written down once, correctly, an hour earlier by someone you pay.
  • And the one that actually hurts: replacing a good rep costs you their ramp twice — once to lose the production they were making, once to pay for somebody new to reach it. Set against that, thirty minutes a job stops looking like a rounding error.
  • Then stretch it. One year is a number you can absorb; five is a number you have to answer for. Turn the model below over and put your own figures in it — including the one that decides whether any of this is real, which is how much of the reclaimed time you would actually refill with work that pays.
A MODEL — SET IT TO YOUR NUMBERS
The five-year horizon

The solid is time your stack spends on your behalf. Everything above it, up to the ceiling, is the same time not spent. Turn it over, then move the sliders until it reads like your business rather than ours — the first one is the number the whole thing rests on.

one full-time person0.65 spent1.35 not spent2 apps · 8 jobs/mo — 68 h over 5 yr2 apps · 13 jobs/mo — 110 h over 5 yr2 apps · 18 jobs/mo — 153 h over 5 yr2 apps · 23 jobs/mo — 196 h over 5 yr2 apps · 28 jobs/mo — 238 h over 5 yr2 apps · 33 jobs/mo — 281 h over 5 yr2 apps · 38 jobs/mo — 323 h over 5 yr3 apps · 8 jobs/mo — 136 h over 5 yr3 apps · 13 jobs/mo — 221 h over 5 yr3 apps · 18 jobs/mo — 306 h over 5 yr3 apps · 23 jobs/mo — 391 h over 5 yr3 apps · 28 jobs/mo — 476 h over 5 yr3 apps · 33 jobs/mo — 561 h over 5 yr3 apps · 38 jobs/mo — 646 h over 5 yr4 apps · 8 jobs/mo — 204 h over 5 yr4 apps · 13 jobs/mo — 332 h over 5 yr4 apps · 18 jobs/mo — 459 h over 5 yr4 apps · 23 jobs/mo — 587 h over 5 yr4 apps · 28 jobs/mo — 714 h over 5 yr4 apps · 33 jobs/mo — 842 h over 5 yr4 apps · 38 jobs/mo — 969 h over 5 yr5 apps · 8 jobs/mo — 272 h over 5 yr5 apps · 13 jobs/mo — 442 h over 5 yr5 apps · 18 jobs/mo — 612 h over 5 yr5 apps · 23 jobs/mo — 782 h over 5 yr5 apps · 28 jobs/mo — 952 h over 5 yr5 apps · 33 jobs/mo — 1,122 h over 5 yr5 apps · 38 jobs/mo — 1,292 h over 5 yr5 jobs/mo40 jobs/mo5 apps1 app — nothing to re-enter
drag or arrow keys to turn · click then scroll to zoom · hover a panel for its numbers
At 25 jobs a month on a 4-app stack, re-entry costs 638 hours over 5 years 0.31 person-years. Refill 50% of that with work that pays and it is worth $31,875.
The first two shape the solid, because they are time. The last two only price it — moving them must not change a shape that means hours, or the chart would be arguing that time is money by construction.
What this is and is not. Every number that carries weight here is a slider, including the one everything rests on: 30 minutes of re-entry per job on a four-app stack. That starts at the figure this piece uses, but it is a starting point rather than a finding — trace one real job with a pen and set it to what you actually find. The shape across stack sizes holds either way: one system costs 3 re-entries a job, and each seam between apps adds about 5.7 more, straight-line rather than exponential, because facts travel along a chain and not between every possible pair of apps.

The dollar figure is the softest number here and is meant to be. Reclaimed time is capacity, not revenue — it becomes money only if you refill it with work that pays, which is why that slider does not start at 100 and probably should not sit there. And the multiple between a four-app stack and one system applies to the paperwork around a job, not to the job. Nobody roofs a house six times faster. The honest claim is hours back, not houses.

The Friday night

Every owner in this trade knows this room. It is nine or ten at night, the house has gone quiet, and you are at the kitchen table with a laptop and a stack of photos. You are not selling. You are not building anything. You are reconciling — working out what a rep actually closed so you can work out what to pay him, finding the photos for a supplement that was due Tuesday, matching an invoice to a job because two of them have the same street name. It is not hard work. That is what makes it corrosive. It is an hour of your life spent being a database, and you will do it again next Friday, and the Friday after that, and nobody will ever thank you for it because nobody will ever know it happened.

Why hiring makes it worse

This is the trap, and it is the reason the ceiling does not move when you add people. Friction does not divide across a team. It multiplies. Every new rep is another person entering the same facts into the same disconnected places, another set of photos in another camera roll, another version of the truth about where a job stands. Two reps can hold the state of the business in their heads and reconcile it over a beer. Six cannot. So the office grows to absorb it, and now you have hired somebody whose actual job — whatever the title says — is to be the integration layer between tools that do not talk. That hire feels like scaling. It is the opposite: it is paying a salary to keep a workflow problem invisible for another year.

  • The tell is simple. If adding a rep requires adding office capacity, you are not scaling a business. You are scaling a workaround.

What to do Monday

None of this needs a purchase order. Do it in the order below, and notice that the first three cost nothing but attention.

  • Trace one job end to end and count the re-entries. One job, one pen. You now have a real number instead of a feeling.
  • Ask your best person what they do after hours that they have never mentioned. Ask it plainly and do not defend the answer. What comes back is the friction they have been absorbing on your behalf, and it is the most accurate audit you will ever get.
  • Find the one fact that gets re-entered the most. It is usually the address or the measurements. Fix that single path first — one fact travelling cleanly beats five travelling badly.
  • Only then look at tools, and judge them on taps removed rather than features added. A tool that adds a capability and three taps has made your Friday worse.
  • Ask the same question of us. Curbsight exists because the job file should be born carrying what the rep already knew at the door — but the right test is not our feature list, it is whether your closer gets his evening back. Bring one real job to a walkthrough and count the taps against your current stack.

Frequently asked questions

How much time do roofing companies actually lose to admin?
Around thirty minutes per job of pure re-entry is a conservative starting point — measurements re-typed, photos re-found, customers re-entered, job stages chased by text. At twenty-five jobs a month that is roughly a hundred and fifty hours a year, just under four working weeks. Trace one real job with a pen rather than trusting the estimate; most owners find their own number is higher.
Why does hiring more people not fix the bottleneck?
Because workflow friction multiplies with headcount instead of dividing. Every additional rep enters the same facts into the same disconnected places and holds another private version of where a job stands. Two people can reconcile that between themselves; six cannot, so the office grows to absorb it. If adding a rep means adding office capacity, the constraint is the workflow, not the staffing.
How do I know if my software is costing me money?
Count taps rather than hours — every time a person re-enters a fact somebody on your payroll already knew. Then ask your best employee what they do after hours that they have never mentioned. A good team absorbs friction silently rather than dropping balls, so the cost never appears as a lost job. It appears as their evenings, and eventually as their resignation.
What is the real cost of disconnected roofing software?
Three costs, and only the first is obvious. The hours, which you can compute. The owner-hours inside them, which are the most expensive on the payroll and get spent on data entry. And turnover — replacing a strong rep costs their ramp twice, once in lost production and once in paying somebody new to reach it. The third usually exceeds the first two combined and is almost never attributed to the workflow that caused it.
About the author

Sol Zendejas-SmithSol 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.

MORE LIKE THIS

Get the next one.

Buying guides, cost breakdowns, and storm-season playbooks for home-services owners — written the way this one was. No cadence promises and no filler: we send when there is something worth reading, and every email has a one-click unsubscribe.

Ready to see your scored territory?

We'll build a sample territory for your service area, walk you through the platform, and answer questions on a 30-minute call. The walkthrough is free.
BOOK A DEMO