The problem
The Netherlands Enterprise Agency (RVO) runs hundreds of schemes for entrepreneurs, from subsidies to permits and advice. But knowledge about those entrepreneurs was spread across dozens of systems, each with its own definitions, labels and consents. Every question that cut across schemes, such as which parties in an area are involved in a heat network, or which entrepreneurs are not yet being reached, took more than a year of manual work and could not be repeated afterwards. RVO wanted to stand beside the entrepreneur as one coherent service provider, but was forced to steer on numbers of applications rather than on results. For seven years, attempts were made to solve this. Each time they hit the same walls: no common language between systems, no reliable way to establish that two records concern the same entrepreneur, and the fear of a megalomaniac replacement project.
The solution
The customer data foundation we built leaves the existing systems in place and connects them with a shared layer of meaning and access on top. The difference this time was AI. Where earlier attempts ran aground on manual harmonisation, an AI connecting layer now standardises scheme names and company identification across all sources and matches customers who are recorded slightly differently in each system. On top of that we are building a customer view that spans schemes, with a privacy architecture in which consent is recorded properly once. The foundation also makes visible which data quality needs to be improved in the source systems themselves. We did this together with RVO's programme team, from board vision to working technology, starting with the story and the use cases and then the proof in practice.
The result
After seven years of attempts, matching and combining customer data across source systems now works, and is repeatable. The proof of concept has shown this, and the foundation is now being extended and put to use for the first use cases. For the first time, RVO sees the entrepreneur as a whole rather than per scheme. Analyses that used to take more than a year become repeatable. And CRM, customer contact and policy analysis can build on the same foundation, so customer data is recorded properly once and reused across RVO.
