I discovered that the data compression algorithm that PBI uses has a bug. The algorithm searches for repeating data in a column and stores a

I discovered that the data compression algorithm that PBI uses has a bug. The algorithm searches for repeating data in a column and stores a single value. However, for string data it ignores case. So, if your data contains both strings 'my data' and 'My Data', it will arbitrarily store only one of those strings and discard the other. When your data is viewed in the report, the case of the data will be different than what is in the source. Some data sources (like SQL Server) ignore case, by default, when doing comparisons. This does not mean that two strings with different case should always be considered as the exact same data. PBI should consider strings that have the same characters, but have different case, as two different values and not arbitrarily remove one of them.