"Due to the complexity of both the data and the process, this was a challenging project on many levels. First, there was a desire to simplify the payment model based on large amounts of complex data. After careful examination, it appeared that the simpler the model (fewer variables, fewer ‘products’), the more important relationships were lost, which in turn increased the risk of opportunism. We often hear ‘everything is different in healthcare’ as an excuse not to look at something too critically or commercially. But in this case, it was actually true: a community nurse doesn’t look at just five variables, but perhaps a thousand (does the patient have shaky hands, can they make coffee by themselves, do they seem confused or uncertain, what kind of informal care do they have?). You might say that with a limited number of these kinds of traits, you could reasonably predict how much care is necessary. But in practice, there are lots of nuances that nurses consider when determining how much and what type of care is required. The second challenging complexity was at the stakeholder level. We gradually involved a panel of the most important stakeholders in the results. There is a shared interest among stakeholders (providing better care for a reasonable price), but each stakeholder represents a different group, which creates a delicate political situation. Our job wasn’t only to devise a new model – we also had to generate as much support as possible. Considering the diversity and varying individual interests within the stakeholders’ panel, this led to many interesting discussions. All in all, it ended up being a very educational experience."