The group exported to real estate development projects — kitchen cabinets, bedroom wardrobes and bathroom units, with left and right doors, different sizes and different finishes. In practice, between 200 and 500 apartments per project, each with different requirements. Delivery was phased: for a given product, 40 units in phase 1, 37 in phase 2, 59 in phase 3, 0 in phase 4, and so on.
When I arrived in the planning department, one of these projects was already underway.
There was no documentation. No single view of the project. Information existed, but scattered across departments — each with its own dialect, its own spreadsheet, its own understanding of what had been ordered, produced and shipped. No macro view. No traceability.
The questions that should have been simple had no answer: had the factory actually produced what was planned? Had dispatch actually shipped what was recorded? The evidence was handwritten notes on scraps of paper, client documents used as a reference, and items codes entered incorrectly in the billing system. Uncertainty everywhere.
After two weeks I went to the Industrial Director. I told him he had two options: trust the factory and dispatch, or trust the database. Both were wrong. He chose the database. Smart man.
Before touching any tool I stopped and thought about the problem structurally. Could I apply the V-Model here? How would I model this in SysML? Who are the actors? I started by mapping it out for myself — requirements, relationships, flows.
Then: ODBC, SQL, Excel. I adapted the query to return only the orders belonging to the project — items, quantities, phases. And started building a dashboard from that data.
Dashboard and production tracking
The dashboard identified items being produced in excess, items short, and surfaced that information to the factory in real time. Response time to production issues dropped — the information was now in planning and on the factory floor simultaneously.
The security incident
One day I received a call from a junior in the export department. A product needed to be on that day's shipment — which pallet was it on? I checked the database — stock entries had a location field tied to pallet. I found it, answered his question, and shared the file with him so he wouldn't need to call again.
I later found the file had been sent to the client. The server was only accessible via intranet, so no data left the network. But it was enough of a lesson. I immediately created two versions: one for internal use with no restrictions, and one for sharing, with protected sheets and limited visibility.
Validation layer
The tool eventually validated delivery notes and orders associated with each project phase — cross-referencing what had been invoiced against what had been planned and shipped.
Weekly patches
Each week a new version was released with a brief changelog — sometimes a small fix, sometimes a significant addition.
Technical documentation
Nobody asked for it. When I left, I left behind a 50-page technical document covering the V-Model and SysML applied to the problem, the SQL queries, the Excel code, query execution times and server impact, security considerations, and the minimum requirements to run the file.
A project that had no single source of truth now had one. Production errors became visible and actionable in real time. The export department had pallet-level traceability. Phase-by-phase delivery validation was in place. A technical document existed for whoever came next.
The starting point was two weeks of uncertainty with no baseline — some of that initial error was never fully resolved, only managed forward from the decision to trust the database. The file sharing incident was a reminder that access control cannot be an afterthought, even on internal tools. The prototype ceiling remains: Excel and ODBC, functional but not a permanent architecture.
The 50-page technical document was an attempt to leave something that could be built on properly. The queries, the data model, the security requirements — it is all there. What it would take is the decision to invest in a real implementation.