"Data-driven" sounds like something for large corporations with dedicated analytics teams and million-euro budgets. But the truth is that small and medium manufacturers often have the most to gain — and the lowest barriers to getting started. You already generate valuable data every day. The question is whether you are using it.
In this article I share practical steps for manufacturers with 10 to 200 employees. No Big Data buzzwords, no expensive platforms — just concrete ways to go from gut feeling and spreadsheets to a clear, visual overview that helps you make better decisions.
What "data-driven" actually means at your scale
Data-driven does not mean you need a data lake, a team of data scientists, or a cloud subscription that costs more than your best CNC machine. At its core, it means one thing: replacing assumptions with measurements.
How long did the packaging line actually run yesterday? What percentage of parts passed quality inspection this week? Which product has the highest scrap rate? If you answer these questions with "about" or "I think", you are making decisions on intuition. If you answer with a number and a trend, you are data-driven. That is the only difference, and it is a powerful one.
The good news is that most of the data already exists. It lives in your ERP system, on the PLC screens, in the quality logs that operators fill out, and in the maintenance records. It just is not connected or visualised in a way that supports quick decisions.
The data you already have (and probably ignore)
Machine data
Run times, stop reasons, cycle counts, temperatures, and pressures. Your PLCs and machines log this continuously — most of it is just never extracted.
Quality records
Inspection results, reject reasons, rework counts. Often tracked on paper or in a standalone spreadsheet that nobody reviews after the shift ends.
ERP and order data
Order volumes, delivery dates, material costs, customer complaints. Your ERP system has years of historical data — the reports just are not set up to answer the right questions.
From spreadsheet to dashboard
Spreadsheets are where most manufacturers start, and there is nothing wrong with that. The problem starts when the spreadsheet becomes the system: dozens of tabs, manual data entry every morning, formulas that break when someone adds a row, and no one quite remembers which version is current.
A dashboard does not replace the spreadsheet — it replaces the manual effort of turning data into insight. It pulls data from your existing sources (ERP, PLC, quality log), updates automatically, and shows the current state at a glance. No opening files, no waiting for Excel to calculate, no arguing about which number is right.
The transition does not have to be dramatic. Start with one screen that shows three things: today's production count, current machine status, and the top reject reason. That single view, updated automatically, is already more powerful than most weekly reports. You can always add more later.
Five KPIs that matter for small manufacturers
You can track hundreds of metrics, but starting with five proven KPIs covers most decision-making needs. These are the ones I recommend to every manufacturer I work with:
OEE — Overall Equipment Effectiveness
Combines availability, performance, and quality into one number (0-100%). It tells you how much of your theoretical capacity you are actually using. Most manufacturers score between 40-60% — there is almost always room to improve.
First Pass Yield
The percentage of products that pass quality inspection without rework. Drops in yield often signal a tooling issue, a material problem, or a process change that went unnoticed.
Unplanned Downtime
How many hours per week your machines stand still unexpectedly. Track the reason for each stop — patterns emerge quickly and point you to the machines or components that need attention.
On-Time Delivery
The percentage of orders shipped on or before the promised date. If this number slips, the cause is usually hidden in one of the other KPIs — low OEE, high downtime, or quality issues that cause delays.
Energy per Unit
Energy cost divided by units produced. As energy prices rise, this metric becomes increasingly important. It also reveals inefficiencies — machines idling, compressors running at partial load, or heating cycles that are too long.
Getting started: a practical three-step approach
Week 1: Pick one question
- Choose the single most important question you cannot answer today
- Identify where the data lives (ERP, PLC, paper log)
- Set up a simple automatic data export or connection
- Build one dashboard panel that answers that question
Month 1: Expand and validate
- Add 2-3 more KPIs to the dashboard
- Share it with operators and managers — get feedback
- Fix data quality issues that surface
- Make the dashboard available on a screen on the production floor
After the first month you will already see patterns you never noticed before. A machine that stops for 15 minutes every shift change. A product variant with twice the reject rate of others. A day of the week with consistently lower output. These are not big data insights — they are obvious facts that were invisible because no one was looking.
When to bring in help
You can get surprisingly far with basic tools and a bit of curiosity. But there are moments when external expertise saves weeks of trial and error: when you need to read data from PLCs that do not have an easy export, when you want to combine data from multiple systems into a single view, or when the dashboard needs to work reliably for years without manual intervention.
That is where an engineer who understands both the factory floor and the software side can make the difference. Not a consulting firm that delivers a 100-page report, but someone who sits next to your operators, understands the machines, and builds something that actually gets used.
Curious what data-driven decision making could look like for your production? I help small manufacturers in the Achterhoek and beyond set up practical, low-maintenance dashboards that connect to existing machines and systems. Let's have a conversation — the first call is always free.