From Excel sprawl to an AI roadmap

The challenge

Our client, a global industrial manufacturer, operates a network of factories and service locations across several European countries. Like most established industrial operations, their commercial backbone runs on processes that have grown over decades: offers are calculated, orders are processed and projects are managed with a mix of ERP systems, document management tools and, above all, Excel.

The leadership team knew AI could make these processes faster and more consistent. What they lacked was a grounded answer to the questions that matter before any tool gets built: where would AI actually pay off, what should be standardized first, and which use case is worth implementing immediately?

Our approach: workshops before prototypes

TransFactor's engagement model is deliberately workshop-first. Instead of starting with technology, we started with the people who run the processes. Together with process owners from the client's locations, we mapped the offer-to-order-to-project flow, collected the real working documents behind it, and scored AI opportunities by impact and feasibility.

The engagement ran in five steps:

1. Discovery workshops with process owners across the factory network
2. AI-assisted analysis of the actual working documents and tools behind each process
3. Prioritization of high-impact AI opportunities, each with a return-on-investment assessment
4. An AI/digitalization roadmap prepared for stakeholder approval, with process standardization recommendations
5. One proof of concept specified in full detail, ready for immediate implementation

What AI-assisted analysis surfaced

One example of why the document-level analysis matters: a single calculation workbook used in the quotation process turned out to exist in multiple parallel variants for different process stages. Each variant contained over twenty interlinked worksheets with cascading formula chains and shared lookup tables, maintained by hand, with change histories that had quietly drifted apart between the variants.

Using AI tooling, we mapped the full structure, formula dependencies and exact differences between the variants in hours rather than the weeks a manual review would take. Findings like these did two things at once: they gave the standardization recommendations a concrete, evidence-based foundation, and they demonstrated to the client's team, on their own documents, what AI-assisted process analysis can do.

Why workshop-first works

AI initiatives in industrial operations fail most often for non-technical reasons: the use case does not match how work actually happens, or the people who would use the tool were never involved in choosing it. Running discovery as workshops with process owners, and validating opportunities against real working documents rather than idealized process diagrams, removes both failure modes before a single prototype is built.