Field Service Automation: The Complete Guide for Trades Businesses

The back office nobody sold you

You bought the good software. ServiceTitan, maybe. Or Jobber, or Housecall Pro, or Service Fusion. It runs your day fine: calls, dispatch, tech notes, photos of the unit before and after.

QuickBooks runs the money.

And stretched between the two: a dispatcher’s spreadsheet, a whiteboard photo texted around every morning, and a downloads folder full of exports nobody has time to open.

You were sold the field side and the money side. Nobody sold you the connective tissue. That’s what this guide is. I’m Bradley Roth, 24 years in IT, and lately most of my work is building exactly that tissue for trades businesses: plumbing, HVAC, electrical, utilities, contractors. What follows is the map I wish every owner had in hand before calling anybody, including me.

The whole guide in one sentence: your office is already automated — in pieces — and the hours leak where the pieces don’t touch.

Where the hours actually leak

Five leaks show up in trades back offices over and over. Any one of them is survivable. Stacked, they eat a whole salary.

1. The dispatch sheet that gets rebuilt every morning

Every day somebody (usually the dispatcher, usually before the phones start) rebuilds the day: who’s on which job, in what order, with what parts. The data already exists in ServiceTitan or Jobber. But pulling it into something the trucks can actually use means exporting, sorting, formatting, printing, texting. Every. Single. Morning.

A dispatcher spending two hours a day on that is spending 520 hours a year — $14,560 at $28 an hour. Not a guess; the math is here, pre-filled. Even if automation only hands back 60% of those hours, conservatively that’s 312 hours a year for one process at one shop.

2. Time cards to payroll

Techs clock in and out in the field system. Payroll runs in something else. So every pay period, somebody walks hours from one screen into another, split by job, checked against the schedule, fixed where somebody forgot to clock out for lunch. It’s careful work, it’s boring work, and it’s exactly the kind of careful, boring, rule-following work software does better than people.

3. Job-costing reports

You want to know what a job actually made: labor, materials, equipment, comebacks. The ingredients live in three or four places, so the report gets assembled by hand — or doesn’t get assembled at all, which is worse. Owners who can’t see job costs are flying on feel, and feel is expensive at bid time. I wrote a whole piece about getting these reports off manual assembly, and the case where a CFO’s six-hour Monday build became five minutes at 5 AM.

4. Work-in-progress tracking

Jobs that are done but not invoiced. Invoices that are sent but not paid. Every trades office has a WIP number that’s “roughly right,” and roughly right quietly becomes “we invoiced $30K late.” Tracking it by hand means someone reconciling job status against invoices against payments: a spreadsheet maze with a shelf life measured in days.

5. Invoice re-keying

The big one. The field system creates the invoice. QuickBooks needs the invoice. And between the two sits a person retyping: amounts, customer names, job numbers, line items. I priced the whole copy-paste loop in this article: a modest 90 minutes a day of re-keying costs a shop $8,580 a year in wages before you count a single mistyped invoice.

What to automate first: frequency × pain × data cleanliness

You can’t fix everything at once, and you shouldn’t try. Score every manual process on three axes:

• Frequency. Daily beats weekly beats monthly. A process that runs 260 times a year has 260 chances to leak hours and create errors. Frequency is the multiplier on everything else.
• Pain. Whose day does it wreck? A process that delays invoicing or blocks payroll counts more than one that mildly annoys the owner. Pain is where the momentum to actually change comes from.
• Data cleanliness. This is the axis people skip. Automation is a mirror: if job data in ServiceTitan is inconsistent (three spellings of the same customer, techs typing “parts” on every line), the automation surfaces that immediately. A process with decent data at both ends is a fast win. A mess needs a cleanup pass first, and it’s better to know that on day one.

The sweet spot: high frequency, real pain, tolerable data. In trades offices that’s almost always some flavor of the invoice loop, the dispatch sheet, or the weekly report. Start there, bank the hours, then reinvest them in the messier stuff.

One more rule: automate the thing you can measure. Before touching anything, know roughly how long the manual version takes per week. If you don’t, run it through the calculator first; vague processes make vague projects.

Build it, Zap it, or have it built: the honest version

There are three real ways to connect systems, and each has a section of town where it’s the right answer.

Do it in the software you have. Some of the leak is a settings problem: ServiceTitan and Jobber both do more than their owners use. Before buying or building anything, make somebody who knows the platform audit what’s already paid for. Sometimes this fixes a leak for free. Sometimes it doesn’t, and now you know.

Middleware: Zapier, Make, Power Automate. Connectors that move records between systems without code. For simple, low-volume, one-directional flows (a new customer here shows up there), they’re fine. They’re also where I meet a lot of shop owners: a stack of zaps that broke silently, a billing change nobody noticed, a sync that skips anything unexpected instead of yelling about it. Here’s the honest comparison of where these tools stop being the answer, usually around volume, complexity, or the moment the process matters to payroll or billing.

A dedicated build: custom business automation. A script or small service built for your exact loop: bi-directional, error-handled, logging everything, running on a schedule, boring forever after. This is most of what I do. It’s also the most expensive tier on purpose; you buy it when the process is load-bearing: invoicing, payroll inputs, job costs. The case where 400 mailed envelopes became a 30-second script is what this tier looks like when it lands.

What each tier costs, honestly: middleware subscriptions run tens to a couple hundred dollars a month, plus your setup time; a dedicated build starts around $7,000, most land around $13,500, and large multi-month builds have reached $35,000. The pricing page lays out the bands and what moves a number up or down. If reading that still doesn’t tell you which tier your problem lives in, that’s a fine thing to ask me directly; I’d rather talk you out of the expensive option than into it.

The rest of this hub

This page is the hub. The spokes, each one a deeper dive:

• Automate your weekly report in Excel: the honest ladder from formulas to a real build, and when to stop fighting Excel
• Stop using Excel as a database: five signs the spreadsheet is costing real money
• The real cost of re-keying data between systems: copy-paste hell, priced
• ServiceTitan to QuickBooks: why nothing talks to each other, and three ways to fix it
• Zapier vs Make vs Power Automate vs a custom script: where the duct tape holds and where it lets go
• What custom business automation costs for a small business: budget framing without the fog
• Automated report distribution: every manager gets their numbers at 5 AM
• Automating document distribution: from 400 envelopes to 30 seconds
• PDF invoices to raw data: a case study where version one failed, and what replaced it
• ERP to CRM integration in plain English: what bi-directional sync means, and the per-field rule that keeps it honest

And the two case studies behind the numbers: the CFO’s Monday and the utility’s mail room.

If one of those describes your week, that’s the one to read next.

Questions I get asked

We run six trucks. Are we too small to automate anything?

No. Truck count matters less than repetition. A six-truck shop with a dispatcher who rebuilds the same daily job sheet by hand has more to gain than a 40-truck shop that already cleaned that up. The honest cutoff is different: if a process happens weekly or more, and a person follows the same steps every time, it's a candidate. If it happens twice a year, keep doing it by hand.

Do we have to replace ServiceTitan, Jobber, or QuickBooks for this to work?

Almost never. The software you bought is good at its job. The problem is the seams between systems, and the fix lives in the seams. I've built sync and reporting that connects the exact stack you already run. Replacement is the last resort, not the plan.

What should we automate first?

Score each manual process on three things: how often it repeats, how much it hurts, and how clean the data is at both ends. High frequency, real pain, decent data: start there. That's usually invoice or job-data re-keying, or the report somebody rebuilds every week. The full decision logic is in the guide above.

What does field service automation actually cost?

Depends on the fix. Middleware subscriptions run tens to low hundreds a month plus setup. A dedicated custom business automation build starts around $7,000, most land around $13,500, and large multi-month builds have reached $35,000. The pricing page spells out the bands and what moves a project up or down.