What Mid-Market Operations Teams Get Wrong About Automation

Most mid-market operations teams come to us after a failed automation attempt. They bought a tool — usually something like Zapier, Power Automate, or a module inside their ERP — built a few workflows, and then watched it quietly stop being used within six months. When we dig into what happened, the root cause is almost always the same: they automated the symptom, not the problem.

The wrong starting point

The instinct is to look at what’s taking the most time and automate that. It sounds logical. But time spent is a lagging indicator — it tells you where pain is concentrated, not why it exists.

We worked with a 12-person operations team at an industrial manufacturer who were losing over 230 hours a month to manual data routing. The obvious fix would have been to automate the copy-paste. But when we mapped their actual workflows, we found the real problem: four separate systems that held overlapping data with no single source of truth. Any automation built on top of that would have been fragile — automating the wrong data, in the wrong format, at the wrong time.

The question that changes the outcome is not “what should we automate?” but “why does this manual work exist at all?”

What a real automation engagement looks like

Before writing a single workflow, we spend time with the people doing the manual work. Not their managers — the people whose hands are on the keyboard. They know exactly where the edge cases are, which steps can’t actually be skipped, and which workarounds have become load-bearing.

In that manufacturer engagement, the operations manager had become a human integration layer — manually reconciling data between their production management system and SAP Business One because nobody had ever connected them. Once we understood that, the automation design became obvious: build the integration first, then layer the workflows on top.

We ended up deploying 23 n8n workflows connecting their production system, SAP B1, Gmail, Slack, Google Sheets, and four supplier APIs. Purchase order processing dropped from 4.2 hours to 18 minutes. Zero missed material shortage alerts in the months following go-live. $190,000 in annual labour cost recovered.

None of that would have happened if we’d started by automating the copy-paste.

Where most tools get it wrong

Off-the-shelf automation platforms are built for the happy path. They handle clean inputs, predictable outputs, and simple conditionals well. Real operations workflows are none of those things.

Supplier emails don’t arrive in a consistent format. ERP exports have unexpected nulls. Approval chains have exceptions that exist because of a decision made three years ago that nobody remembers the reason for.

This is why we almost always use self-hosted, open-source tooling like n8n for production workflows — it gives us the flexibility to handle complexity without hitting a wall every time real-world messiness appears. And because it runs on your infrastructure, the data stays inside your environment.

The engagements that stick

Automation that lasts has three things in common. First, it was designed around the process as it actually works, not as it’s supposed to work on paper. Second, it was built with the operations team, not delivered to them — they need to understand what’s running and why. Third, it has monitoring built in from day one. Silent failures are the fastest way to erode trust in automated systems.

The teams that get the most from automation are not the ones who automate the most things. They’re the ones who automate the right things, completely, and then leave them alone.

If you’re evaluating automation opportunities in your operations and want a second opinion on where to start — book a call. We’ll map it out before recommending anything.