The problem isn't the spreadsheet. It's the costs hidden inside it.
Dutch construction companies spend an average of 11 hours per week per procurement officer on manual comparison work. This is what that means.
What it costs in time
11 hours
per week per procurement officer on comparison work
Remarcable / BuildMate 2025
Three procurement officers. 46 working weeks. €65 per hour. That's €148,500 per year in unproductive time, for supplier comparison alone. Not including the cost of the wrong decision.
That decision also costs money. In an internal Skanska experiment, a generic enterprise AI system produced errors of €8,354 on a single tender line. Manual comparison in Excel offers no guarantee. Neither does a system that generates answers.
What happens when the procurement officer leaves
Most procurement knowledge isn't in the system. It's in the head of the procurement officer who has known for twenty years which supplier is reliable, which certifications hold up and which quote looks good but isn't.
Eight out of ten Dutch contractors already face staff shortages. When that procurement officer leaves, the knowledge leaves with them. There is no system that captures it.
NL Times / Euroconstruct 2025
Regulation is making it more urgent
The documentation obligation already existed. But three developments make it concrete and enforceable now:
- Aanbestedingswet: every supplier decision by a municipality must be documented, consistent and defensible before the ACM.
- CPR 2024/3110: new certification requirements for construction products take effect from January 2026. The obligation exists. Most organisations lack the tools to comply.
- CSRD Scope 3: increasing reporting requirements for supply chain emissions require structured supplier data.
The spreadsheet is no longer a legally adequate answer.
Generic AI doesn't solve this
€8,354
error on a single tender line when using a generic enterprise AI tool
In an internal Skanska experiment, a generic enterprise AI system produced different numerical outputs for identical procurement inputs. This is not an edge case. It is the current state of generative AI in procurement: the system generates an answer, but cannot reproduce the calculation.
Hamppi 2025, Aalto University. Case study based on internal Skanska experiment.
A system that generates answers is not the same as a system that calculates them. Dundir uses a mathematical optimisation engine. Every recommendation is deterministic, reproducible, and comes with a full reasoning trail.
Curious what manual procurement costs your organisation?
Take the scanFive questions. Specific to your team size and volume. No email required to see the result.
Curious what this costs in your organisation? The scan gives a personalised estimate in five questions.
No email required to see the result.