A collection of keys must be defined for each dimension attribute. This key collection must uniquely define each member of that attribute.
If we take a date dimension for example, the key for the [Year] level can be defined as the [Year] column in the database. Every member with the same year will be aggregated together. The story is different once we get to the [Quarter] level. We have many members named [Q1] and they become unique in conjunction with the [Year] member. That means that the key for the [Quarter] level will be [Year] and then [Quarter]. The same goes for the [Month] level as well. The key would be [Year] and then [Month]. You could use the month name or the month number – whichever you like. At the day level, you have a couple of options. First of all, you can use the identity ID of your dimension table, and this is more desirable when you are using a smart date key or you can choose to use [Year], [Month], and [Day] as the key.
Now every member of each attribute and consequently any hierarchies built using these attributes has a unique way to reference each member.
After you are finished, you should make sure that your attribute hierarchies are set up correctly or SSAS won’t design [
can’t calculate] aggregates at any level other than the [All] level and the leaf level. More on that here.