How do you automate the procurement process without losing control?

Summary: Orders for materials and services at technical companies often still run through phone calls, e-mail, and scattered messages. Without structured data, orders lose their link to the project they were meant for, resulting in wrong deliveries and duplicate orders. Automating without oversight makes that risk worse, not better. This article shows how to build a procurement process around standardized product data and a direct link to project budgets, so discrepancies always stay visible.

Last updated: 3 August 2026 · By Jack van der Vall, AI Engineer

Related: also read how to automate purchase invoice processing without losing control, how to link legacy systems to modern automation, and our process automation services.

Why does procurement at technical companies still run on manual work?

Ordering materials for a project sounds simple. In practice, it often runs through a mix of phone calls, e-mail, and scattered messages to the supplier. That is often not recorded centrally or systematically.

Techniek Nederland, the Dutch trade association for the technical sector, describes this problem sharply. According to Techniek Nederland’s review of improving the ordering process, orders are communicated and handled through different channels. As a result, the wrong item can be delivered at the wrong time in the wrong place.

That is not a minor annoyance. A wrongly delivered part on a job site means waiting, reordering, or improvising with what is on hand. At a technical company running multiple projects at once, that adds up.

Yet digitalization of processes like this remains limited. Only 23% of Dutch businesses with ten or more employees used AI technology in 2024, according to figures from Statistics Netherlands (CBS). That makes it plausible that for an operational process like procurement, manual work is still the norm rather than the exception.

What goes wrong if you only automate the order itself?

The reflex is: automate the ordering process. A system that sends orders to suppliers automatically sounds appealing. But that is where a new risk begins.

Automating just the order action does not fix the underlying problem. If the data behind it stays messy, a system spreads that mess faster. A wrong article number is no longer passed on incorrectly once, but continuously.

This pattern is not unique to one company. According to NEVI, the Dutch association for procurement management, international research finds that procurement and contract management professionals often work with outdated systems, where data quality is poor and user adoption stays low.

A system that passes orders through automatically without structuring the underlying data mostly automates the chaos, not the process. The same principle applies here as with invoice processing: automating the action without the verification behind it makes the problem bigger, not smaller.

How do you build a procurement process that keeps oversight?

The solution is not a choice between manual work and blind automation. It is about the order of operations: structure first, automation second.

The Dutch construction and installation sector already has a standardized foundation for this. Ketenstandaard Bouw en Techniek, the Dutch industry body for shared data standards, maintains the ETIM product classification and GS1/GTIN coding, which allow suppliers and articles to be identified unambiguously. ETIM itself is used across European technical wholesale, not just in the Netherlands. The same 16A circuit breaker arrives from three wholesalers under three different descriptions. A system that matches on free text sees three different products. A system that matches on the GTIN code sees one. These codes are not a side detail. They are the anchor point automated matching can rely on.

With that foundation, a system can do three things that scattered messages cannot. It recognizes exactly which article is being ordered, regardless of what the supplier calls it internally. It links the order directly to the project and the available budget. And it flags when an order deviates from that, before it goes out the door.

graph TD
    A[Order request from a project] --> B[Article recognized via ETIM/GTIN code]
    B --> C{Does the order fit the project budget?}
    C -->|Yes| D{Is the supplier on the approved list?}
    C -->|No| E[Order presented to a work planner]
    D -->|Yes| F[Order automatically placed with the supplier]
    D -->|No| E
    F --> G[Delivery checked against the placed order]
    G -->|Matches| H[Order completed]
    G -->|Deviates| E

Accessible summary: A flowchart of controlled procurement automation. An order request from a project is recognized via a standardized product code. The system first checks whether the order fits the project budget, then whether the supplier is on the approved list. If it passes both, the order is placed automatically and later checked against the actual delivery. If budget, supplier, or delivery deviates, the order goes to a work planner.

How do you stop two people from placing the same order?

A separate pitfall: two colleagues independently ordering the same material for the same project. That happens the moment nobody can see what the other has already ordered.

An order that stays invisible to the rest of the team can easily get placed a second time. The solution is the same reference used for matching: link every order to a project, cost code, and requester, and make that link visible to everyone before a new order goes out.

That visibility has a second benefit: post-calculation. If you still know exactly what was ordered for what at the end of a project, you can see precisely where the budget went. Without that link, you have to reconstruct it afterward from scattered delivery notes and invoices.

Where do you start with automation?

The order of operations determines success. Do not start by automating the order action itself.

1. Count how many of last month’s orders you can trace back to a project and cost code, without having to ask anyone. That number shows you how big the gap is.
2. Map your article flow and structure the data first, not the process. Which product codes do you already use, and do they align with a standard such as ETIM or GS1/GTIN? A clean article and supplier list is the foundation automated matching can run on.
3. Link orders to a fixed, team-visible reference: project, cost code, and requester. Without that link, post-calculation stays manual work and duplicate orders slip through.
4. Automate the verification first, then the placing of orders. A system that reliably flags discrepancies earns trust before it is allowed to order on its own.

Raw data available: Procurement Process Automation Source Overview

---

Frequently Asked Questions

Do I need to redesign my entire procurement process before I can automate it?

No. Start by structuring product data using existing classifications such as ETIM and GS1/GTIN. Then link orders to projects and budgets. Only once that foundation is in place should you expand automation to more of the process.

Does automation replace my buyer or work planner?

No. Automation takes over the routine matching and checking. The buyer or work planner reviews the exceptions: orders that fall outside budget, involve an unknown supplier, or deviate from what the project actually needs.

Does this work alongside my current ERP system?

Usually, yes. This approach does not replace your ERP; it structures the data around it and safeguards the link between order, project, and delivery. That makes it feasible even for companies running older or less flexible systems.

---

Key Takeaways

• Manual procurement through phone calls, e-mail, and scattered messages leads to wrong deliveries and orders that lose their link to the project.
• Automating just the order action fixes nothing if the underlying data stays messy: the system ends up automating the chaos.
• Standardized product data, such as ETIM and GS1/GTIN, is the anchor point reliable automated matching depends on.
• Link every order visibly to a project, cost code, and requester: that keeps post-calculation possible and prevents duplicate orders.
• Automate the verification first, then the placing of orders: trust is built through reliable flagging of discrepancies.

---

About the author

Jack van der Vall is an AI Engineer at Opusmatic, specializing in AI automation for technical installation companies and SMEs in Zuid-Holland. He helps installation companies identify and automate time-consuming business processes.

Opusmatic | LinkedIn | Contact