Every contractor we talk to has one person who holds the whole office together. Usually she has a title like office manager or operations coordinator, and sometimes she’s also the owner’s sister. She answers the phone, books the jobs, chases the paperwork, and knows where everything is.
She’s also spending about a quarter of her week retyping things that already exist somewhere else.
Nobody planned this. It accumulated: a form got added here, a spreadsheet there, a new scheduling tool that doesn’t talk to the accounting system. Each addition cost fifteen minutes a day, which is invisible. Add up eight of them and you’ve got most of a workday, every week, spent moving data from one box to another box.
Where the hours actually go
When we run a time audit at a small trades company (typically 8 to 40 employees), the same handful of tasks show up almost every time. These are rough figures from our own engagements, not a survey, but the shape is remarkably consistent.
- Lead intake. A call comes in, or a form submission from the website, or a Facebook message, or an Angi lead. Someone reads it and types the name, address, phone, and problem description into the CRM. 3–5 minutes per lead. At 15 leads a week, that’s an hour.
- Quote assembly. Pull the customer record, copy it into a quote template, retype the line items from the tech’s notes, export a PDF, attach it to an email, write the email. 10–20 minutes per quote.
- Job-to-invoice handoff. The job is done. Now someone opens the job record, copies the customer and line items into QuickBooks, and sends the invoice. 5–10 minutes per job, and it’s where most billing errors are born.
- Scheduling confirmations. Texting or calling customers the day before. 2 minutes each, times every job on the board.
- Weekly reporting. Exporting three CSVs, pasting them into a spreadsheet, fixing the columns, sending the owner a number he’ll glance at for eight seconds. 45–90 minutes every Friday.
- Subcontractor and supplier paperwork. Certificates of insurance, W-9s, PO numbers, all chased by email and filed by hand.
Five hours is a conservative middle; we’ve seen offices where it’s closer to twelve. Argue with the number if you like, but notice what’s underneath it: none of that work requires judgment. It requires a human only because two systems refuse to speak to each other.
Why this is worth fixing before you buy any AI
There is a lot of noise right now about AI for the trades. Most of it is premature. If your customer data lives in four places and none of them agree, an AI assistant will confidently give wrong answers faster than a human would. Clean plumbing first.
The other reason: copy-paste work is the cheapest thing to automate and the easiest to prove. You don’t need a model, you need a webhook and a mapping. A quote-to-invoice sync takes a few days to build and pays back in the first month, which is a much easier conversation with an owner than a six-figure platform migration.
How to run the audit yourself
You don’t need a consultant for the first pass. Do this over one week.
- Have the office manager keep a running note. Every time she copies something from one screen to another, she writes down what it was and roughly how long it took. One line each. No formatting, no spreadsheet, just a note.
- At the end of the week, group the lines. You’ll get five to eight clusters, not fifty. Most offices have fewer distinct tasks than they think.
- For each cluster, write the frequency and the minutes. Multiply. Sort descending.
- For the top three, write down the source system and the destination system by name. Not “the scheduler,” but Jobber. Not “accounting,” but QuickBooks Online.
- Ask one question per cluster: does anything here require a judgment call a stranger couldn’t make from the data alone? If yes, that step stays human. Automate the steps around it.
That last step matters more than the rest. The usual way automation goes wrong isn’t building the wrong integration; it’s trying to automate the one decision that genuinely needs someone who knows the customer.
What we usually build first
Once the audit is done, the first build is almost always intake. Every lead source (web form, phone, Angi, Google Local Services, the inbox) lands in one place, gets normalized into the same fields, gets deduplicated against existing customers, and creates one record. Nobody retypes anything.
We build these on n8n, self-hosted, because the volume is unpredictable and per-task pricing punishes exactly the months when business is good. But the tool matters far less than the mapping. A well-specified intake flow in Zapier beats a badly specified one anywhere else.
Second build is usually the job-to-invoice handoff, because it removes billing errors as a side effect. When the invoice is generated from the same record the tech filled out, the line items match the work and disputes drop, which is worth more than the saved minutes.
Third is reporting. Not a dashboard nobody opens: a Monday morning message with four numbers in it, sent to the owner’s phone. Jobs closed, revenue booked, quotes outstanding, average days to payment. If the owner wants more he can ask, and by then you know what to build.
What this is actually worth
Five hours a week at a fully loaded admin cost of roughly $28 an hour is about $7,300 a year. That alone justifies a small build, but it doesn’t excite anyone.
The real return is elsewhere. Faster quote turnaround wins jobs, because when you’re the first quote in the customer’s inbox you win a materially higher share of them. Fewer billing errors means fewer disputes and shorter payment cycles. And the office manager stops being the single point of failure for the entire business, which means she can take a vacation without the schedule falling apart.
That last one is the argument that actually lands with owners. Nobody buys automation to save $7,300. They buy it because they’re tired of the whole operation depending on one person’s memory.
Where to start this week
Run the note-taking exercise. It costs nothing and takes five days. Then look at your top cluster and ask whether the two systems involved have an API. Almost everything in the trades stack does now: Jobber, ServiceTitan, Housecall Pro, QuickBooks, CompanyCam. If both ends have an API, the work is a mapping problem, and mapping problems are solvable in days rather than quarters.
If you want a second pair of eyes on the audit, that’s the conversation we like having. We won’t talk tooling until the list exists.