Workflow automation for STL small business: saving 20+ hours a week
Zapier is a graveyard for half-built automations. Here's how real workflow automation ships, which processes to pick first, and when to build vs. buy.
Every small business we audit has a Zapier account. Most of them have 40+ Zaps, half of which are broken and nobody noticed. That's the Zapier graveyard. It's a real pattern, and nobody talks about it honestly.
The graveyard happens because Zapier is great for prototyping and terrible for production. A Zap that fires 100 times a day breaks when a field name changes, a rate limit hits, or an API response format shifts. The Zap fails silently. Your operations degrade quietly. You don't notice until a customer complains.
Real workflow automation looks different. It's built with observability, error handling, and retry logic. It runs on a stack you control. It's monitored. When it breaks, you know. When it drifts, you fix it.
The 20-hour calculation is straightforward. Take your five most repeated internal tasks: quote generation, invoice chasing, onboarding handoff, reporting, and status updates. Measure the payroll hours they consume weekly. Across a 15-person services business, that number is routinely 20-40 hours a week. Automate it properly and those hours redirect to billable work or margin.
Which processes to automate first.
High volume and repetitive. If it happens more than 20 times a week and the steps are consistent, automate it. Onboarding, invoicing, reporting, data entry.
Rule-based logic. If the decisions inside the process can be written as rules a human wouldn't argue with, automate it. If the decisions require judgment, don't.
Cross-system handoffs. The copy-paste between your CRM and your accounting tool is the highest-ROI automation most operators can run. It's tedious, error-prone, and gets done by someone billed at $60 an hour who should be doing something else.
St. Louis manufacturing and distribution is the single biggest missed opportunity in the region. Mid-size manufacturers in O'Fallon, Maryland Heights, and the I-270 corridor are running inventory, purchase orders, and shop-floor handoffs on processes that haven't changed since 2005. The automation ROI here is 10x what we see in services businesses.
What automation looks like in practice. A professional services firm we shipped for redirected 28 hours a week. An e-commerce operation redirected 18. A property management company redirected 35. All three paid back the build inside 90 days.
Build vs. buy. Buy for commodity workflows with a good SaaS fit (calendar, email, basic CRM). Build for the idiosyncratic operator logic that's specific to your business. The line is clearer than vendors want you to believe. If the tool bends to your workflow, buy. If your workflow has to bend to the tool, build.
Getting started without disruption. Don't automate a process you don't fully understand. Document it first. Run the automation in parallel with the human version for two weeks. Cut over once the automation matches the human output on the edge cases. Skipping the parallel phase is how operators end up with the Zapier graveyard pattern in a new location.