Data that survives the deal.
Most data programmes in industrial groups fail at the same two moments: Day 1 of a carve-out, and the day the TSA expires. This is a working notebook on what actually holds up — architecture, governance, and the AI use cases worth funding — written from inside a multinational manufacturer rather than from a consulting deck.
The problem, as you actually experience it
You’ve closed. The data separation plan looked fine in the SPA. Eleven months later the seller’s ERP still runs your month-end close, three countries reconcile master data in spreadsheets, and the AI pilot the board asked about has no foundation to sit on.
None of this is a technology problem. It becomes one about six months too late.
Where this comes from
Built three times
Enterprise data architectures designed and delivered from scratch, three times, in multi-hundred-million-euro organisations.
Through the deals
Distributed data teams led across onshore and offshore delivery, through carve-outs and integrations rather than around them.
Both sides of the table
Fifteen years split between consulting delivery and running data on the client side, where the outcome actually lands.