Digitalization in Mid-Sized Companies: If You Can't Do Controlling, You Don't Need AI

Nearly half of mid-sized companies don’t do their own controlling. No contribution margins per customer. No cost center analysis. No reliable margin per order. In the same breath, the next statistic – confirmed by Bitkom surveys: Nearly as many plan to deploy AI.

That’s not a contradiction. It’s a pattern — and I see it in almost every consulting project.

Companies skip stages 1, 2, and 3 of digitalization — and want to jump straight to stage 7. I offer this maturity assessment as part of my ERP consulting for SMEs.

Digital Maturity: Which Stages SMEs Are Skipping

Digitalization isn’t a switch you flip. It’s a maturity process with stages that build on each other. Skip a stage, and you’re building on sand.

Stage 1: Clean Data

Is your master data current? Are customer addresses, product descriptions, supplier contacts correct? If not, every system — whether ERP, CRM, or AI — delivers wrong results. Data quality isn’t a precondition for projects, but it is a fundamental prerequisite for digitalization.

Stage 2: Documented Processes

Do your employees know how the process from customer order to delivery officially works? Or does everyone have their own version? If processes aren’t documented, they can’t be automated — because nobody knows what’s actually supposed to be automated.

Stage 3: Functioning Controlling

Do you know your margins per product, per customer, per order? Do you know which processes cost how much? Most mid-sized companies I advise have a P&L and a trial balance. But real controlling that supports operational decisions? Very few have that.

Stage 4: Automation

Only when data is correct and processes are documented does automation make sense. Automatic order suggestions, automatic dunning runs, automatic invoice approval — these only work when the basics are right.

Stages 5–7: AI, Predictive Analytics, Autonomous Processes

AI isn’t stage 1. AI is stage 5 or higher. Germany’s BMWK (Federal Ministry for Economic Affairs) also emphasizes the importance of digital fundamentals in its AI strategy. It requires clean data, defined processes, functioning controlling, and initial automation. Layering AI on top of missing fundamentals automates chaos.

AI Projects Without Controlling: Three Real-World Examples

AI Without Data Quality: The Hallucinating Assistant

A trading company introduces AI-powered demand planning. The model should predict demand and generate order suggestions. Problem: The product master data in the ERP is outdated — wrong units, duplicate item numbers, missing product groups. The AI “learns” from flawed data and delivers systematically wrong forecasts. The result: more wrong orders than before. The project manager blames the AI. The real problem sits in master data management.

Automation Without Process Documentation: Automated Chaos

A service company wants to automate its quotation process. From customer inquiry to finished quote, everything should run digitally and automatically. Problem: The quotation process isn’t defined. Every sales rep does it differently — different calculations, different templates, different approval levels. The automation maps one employee’s process. The other four continue as before. The result: five parallel processes, one of them automated. Efficiency gain: zero.

Dashboards Without Controlling: Pretty Pictures Without Substance

A mid-sized manufacturing company invests in a BI tool. Dashboards are built, KPIs visualized, management reports automated. Problem: The underlying data is wrong. Contribution margins are calculated incorrectly because cost centers aren’t properly assigned. Cycle times are incomplete because ERP confirmations aren’t consistently posted. The result: management makes decisions based on wrong numbers — and trusts them because they come from a professional dashboard. That’s more dangerous than no dashboard at all.

The Maturity Check: Five Honest Questions

Before you start your next digitalization project, answer these five questions honestly:

1. Can you name your highest-revenue product’s margin within 10 minutes?

If not, you’re missing operational-level controlling. That’s nothing to be ashamed of — but it’s the wrong foundation for AI.

2. Do your five most important business processes have a current, written description?

Not the ISO manuals from 2017. A current description showing how the process actually runs today. If not, you can’t meaningfully automate that process.

3. What percentage of your ERP features do you actively use?

Most mid-sized companies use 30–40% of their ERP functions. Before buying AI tools, check whether your existing system has features you’re not even using yet.

4. Do you trust your reports?

If your managers regularly recalculate ERP reports in Excel, you have a trust problem with your data. AI won’t establish that trust — it will undermine it, because it builds on the same unreliable data.

5. Have you made an investment decision based on data analysis in the last 12 months?

If the answer is no, you don’t need better technology. You need a culture that makes data-driven decisions. That’s not a technology topic — it’s leadership.

Maturity Check: Is Your Company Ready for AI?

The solution isn’t to reject AI. AI will fundamentally change how mid-sized companies work. But the sequence determines success or failure.

Fundamentals First

Quarter 1: Check and clean master data — not everything at once, but the 20% that affects 80% of business processes. Simultaneously: document the five most important processes — not perfectly, but honestly.

Quarter 2: Put ERP usage to the test. Which features lie dormant? Automatic order suggestions, dunning runs, workflows — what can you activate without buying a new tool?

Quarter 3: Build initial controlling — contribution margins per product, per customer. Not perfect, but reliable. Based on data that’s now clean.

Then AI

From Quarter 4: Now AI makes sense. On clean data, with documented processes, in the context of functioning controlling. The AI pilot starts integrated into your existing systems — not as an isolated toy.

Conclusion

Nearly half of mid-sized companies don’t do controlling. At the same time, nearly half plan to deploy AI. That’s the reality — across all industries.

Digitalization isn’t a technology problem. It’s a maturity problem. And maturity can’t be skipped. Those who start at stage 1 instead of stage 7 will be further ahead in 12 months than any company that jumped straight to AI — without mastering the basics.

Learn to walk first. Then run.

Next Step

Want to know what stage your company is at — and where the biggest lever lies? I do the maturity check with you and show which steps in which order deliver the most impact.

→ Book a free consultation

→ Or read more first: DATEV Interface Check — Self-Assessment