When does it make sense to duplicate data across different microservices?

I'm building a system where a bunch of microservices all need the same main data. But, they often want it tweaked a bit, like in different formats or with just certain fields. My first idea was to keep all the data in one spot and have services ask for it. That's caused some real headaches though, like services getting too tied together and things slowing down when there are lots of reads. So, now I'm thinking about making copies of some data, maybe even simplifying it for what each service needs. Has anyone tried this approach and found it helpful? Or are there hidden problems I should know about, besides the usual worries about keeping everything consistent?