Werkflow
Trades & back office

36 Zaps Instead of an Office Assistant: How a Trade Business Automates Its Back Office

A 15-person plumbing business runs call-backs, quotes, purchasing, leads and ads on automations the owner built himself. The workflow carries over to any small service business.

Reading time: 6 minAs of: Oct 20264 Sources

In small service businesses the back office often depends on one person. They call people back, send quotes, order material and keep an eye on ads. When they are out, the work that brings in money stays undone. The plumbing and heating business 3R Plumbing & Mechanical in Massachusetts has moved almost all of this work into automations. The owner built them himself; he describes himself as a tradesman, not a computer person.

The starting point

The business has 15 employees and works with ServiceTitan, industry software for jobs, quotes and invoices. According to Zapier, revenue grew from about 300,000 USD in the early years to an expected 3.8 million USDH this year. Before the change, missed calls and leads were lost, jobs were attributed to the wrong campaigns, and Google Ads did not know which clicks led to paid jobs. The classic Zaps regularly exceeded the monthly allowance.

Today 36 next-gen ZapsH run with about 1,400 executions a week. Most were built between late September and mid-October 2026. Zapier does not give time savings in hours.

The workflow

WorkflowAI stepPersonSystem and dataCheck
  1. 01StartHourly: pull calls, jobs, quotes, invoices and memberships from ServiceTitan
  2. 02SystemFill one portal per role: call-backs for the phone team, open quotes for sales, orders for the warehouse
  3. 03SystemWeekdays at 6:30 am: morning email with the day's plan and open call-backs
  4. 04CheckIs a booking still untouched after 20 minutes?
    yes: email to the owner, flagged again after 24 hours
  5. 05PersonPurchasing approves an order; the approval goes straight back to ServiceTitan
  6. 06SystemEvery morning: send real calls, bookings and job revenue back to Google Ads
  7. 07CheckAre all data feeds up to date?
    no: alert email naming the stale feed
  8. 08ResultFridays: weekly report with revenue and ratings per technician

The core is unspectacular: one data source is read on a schedule, and each role gets exactly the slice it needs. The AI sits mainly in the building. The owner connected Claude to Zapier through Zapier MCP. He describes what a workflow should do, Claude builds and publishes it, and he checks the logic and the result.H

Two rules from the case

  • Recurring work as a schedule, not a prompt: If you ask a chatbot the same question every morning, turn it into a scheduled workflow. It is cheaper and returns the same format every time.
  • A watchdog for the data: A separate Zap checks every morning that all feeds are current. Without it, a silent failure only shows up when a customer complains.

What it costs

Zapier bills by task. On annual billing, Professional starts at 19.99 USDP a month and Team at 69 USD, each for the smallest allowance. Paid monthly, they cost 29.99 and 103.50 USD. Since 6 October 2026, agents run in Zapier as an AI step inside a normal Zap. Standard models cost no extra tasks; stronger models cost 1, 3 or 5 tasks per step. A run that uses more than 75 tasks pauses and waits for a person.H At about 1,400 runs a week, the smallest allowance is not enough. Which tier a business needs depends on how many steps each workflow has.

Limits and risks

  • Dependence on one person: The owner built every workflow. Document each Zap with purpose, trigger and owner, or the knowledge again sits with one person.
  • Customer data: Calls, addresses and job data flow through several services. Check a data processing agreement for each.
  • Ad data: Sending job revenue back to Google Ads needs a legal basis and a note in the privacy policy.
  • Transferability: ServiceTitan is rare in Germany. The pattern works with any trade software that connects to Zapier, Make or n8n.
  • Where the numbers come from: All figures come from a Zapier case study and are not independently verified.

How common such workflows are in smaller firms is shown by the October 2026 Eurobarometer: 42 % of SMEs and mid-caps surveyed in the EUU use AI or algorithmic management, most often in customer service. Which platform fits which process is compared in our article on Zapier, Make and n8n.

How to start

  1. Write down the three questions asked in your business every morning, such as open call-backs or today's appointments.
  2. Check whether your trade software connects to Zapier, Make or n8n.
  3. Build the morning email with the day's plan and open call-backs first.
  4. Add a watchdog that warns you when data is stale.
  5. After four weeks, count how many call-backs and leads were left undone and compare with before.

Sources

  1. HZapier: From a garage and a pickup truck to $3.8 million (3R Plumbing), 14 Oct 2026
  2. HZapier: Zapier Agents is now AI by Zapier, 6 Oct 2026
  3. PZapier: Pricing (accessed October 2026)
  4. UEuropean Commission: Flash Eurobarometer on AI and algorithmic management, 14 Oct 2026
More from this section