
Insights · September 2, 2026 · 6 min read
What actually gets automated in a 40-person contracting business
A realistic set of processes, what a week of automation work changes, and what it doesn't.
Take a contracting business with about forty people: two owners, an office of four, five crews, a couple of estimators, and a service tech who does the small jobs. Revenue somewhere between six and twelve million. It has a CRM that one owner set up in 2021 and half the office ignores, QuickBooks, a scheduling app the crews like, and a shared inbox that is, if everyone's honest, the actual system of record.
Nothing about this business is broken. It makes money. But every process in it was designed when there were fifteen people, and the extra twenty-five have been absorbed by adding steps rather than changing them.
This is what a week of automation work looks like inside it. Not a transformation. A week.
Monday: what actually happens to a lead
A lead arrives one of five ways: the web form, a phone call to the office, an email to the owner, a referral text to whichever estimator the referrer knows, or a marketplace platform that sends its own notifications. Each of those lands somewhere different. The office manager checks the form inbox twice a day.
The owner forwards emails when he remembers. The estimator with the referral texts back from his truck and nothing gets written down until he's back at a desk.
The map of this takes an hour to draw and about forty minutes of the office manager's time to correct. What it shows is that the median lead waits four hours for an acknowledgement, and about one in eight is never entered into the CRM at all. Those aren't lost jobs, necessarily.
But they are jobs nobody can see.
The automation here is unglamorous. Every channel feeds one intake. Every lead gets an acknowledgement within a minute that says a real person will call, and names the person.
Every lead is assigned by service area and job type, with a follow-up task that escalates if nothing happens in a day. The CRM becomes the record because the record is now created before anyone has a chance to skip it.
Hours back: about four a week for the office, spread thinly enough that nobody notices it as time, and a lead list that's finally complete. The second effect matters more than the first.
Tuesday and Wednesday: quoting
Estimating is where the skill is, and nobody's automating the walk-through or the judgement about what's behind the wall. What gets automated is everything around it. In this business, an estimate is built in a spreadsheet the senior estimator made years ago, copied from the last similar job, adjusted, then re-typed into a Word template because the spreadsheet doesn't look professional, then saved as a PDF, then emailed, then, if accepted, re-typed a third time into the CRM and a fourth into QuickBooks.
Four entries of the same numbers. Each one a chance for a typo, each one twenty minutes. The automation is a quote builder that reads from a maintained price book, pulls the last three similar jobs for reference, generates the customer-facing document from the same data, and writes the accepted quote into the CRM and the accounting system without anyone typing it again.
The estimator still decides every number. He just types each one once.
This is the biggest single win in most contracting businesses. Hours back: six to eight a week across two estimators, and the quotes go out the same day instead of the next.
Thursday: scheduling and the phone
The scheduling app works. The problem is what happens around it. When a job moves, the office calls the customer, texts the crew lead, and updates the calendar, in that order, and if the customer doesn't pick up the whole chain stalls.
Confirmations the day before are done by hand, from a list, between other things.
What gets built: when a job is created or moved, the customer gets a confirmation with the crew lead's name and a window; the crew gets the updated day sheet; a reminder goes out the afternoon before with a reply-to-reschedule option that lands back in the office as a task. None of this is clever. All of it is currently a person's afternoon.
Hours back: three or four a week, and a no-show rate that drops for reasons nobody has to think about.
Friday: money
Invoices are sent when the office manager gets to them, usually within a few days of completion. Follow-up on unpaid invoices happens when the owner looks at the aging report, which is when cash feels tight. The average job pays in forty-one days on thirty-day terms.
The automation: the invoice is generated the moment the crew lead marks the job complete and the office confirms it, which is a one-tap step. Reminders go at fourteen, twenty-eight, and thirty-five days, in the owner's voice, and stop the moment payment lands. What the owner gets on Monday morning is not an aging report; it's a list of the three accounts that actually need a phone call.
Hours back: two or three a week. Days-to-paid usually drops by a week within a quarter, which is worth more than the hours.
What the week doesn't change
The estimator's judgement. The crew's craft. The owner's relationships.
The fact that the business needs a better price book and someone to own it. The automations expose that need; they don't fill it. And they don't fix a process that's genuinely wrong; they make the process as it is run without people carrying it.
The honest total for this business is somewhere between fifteen and twenty hours a week returned to the office and the estimators, from four builds, most of it in the first two. That's not a headcount saved. It's the office manager not working Saturdays, and the second owner getting his estimating time back.
Which is what they wanted when they asked.
If you're this business
- Start with intake. It's the cheapest build and it makes every other number visible.
- Do quoting second. It's the biggest win and it forces the price book to exist.
- Leave the CRM argument alone until the intake is automatic; the CRM gets used when the data is already in it.
- Measure days-to-paid before and after. It's the number that convinces the skeptical partner.
The AI Workflow Audit does the map and the first two builds in five days. The free Workflow Finder will tell you in three minutes which two those are likely to be.
Automation · Trades · 01 Software & AI Systems
The AI Workflow Audit takes 5 days and starts with a map of exactly this. Read the offer, or try the free Workflow Finder.