• Cool. Thanks for that.

    I have done some tests since confirming that, so char and nchar show much better savings, and int's dependant on the size of the values show a lot of savings also.

    But, in most applications i know, text is stored as varchar, such as in address tables etc, and this is where i thought sparse columns could help a lot as address data can be very variable, i.e some people have postal codes, some dont, some dont have a 3rd address line etc.

    It seems though that there is no benefit of adding sparse property to varchar fields, and there may actually be a detrimental effect.

    You say that there is little difference due to the way variable length col's store nulls. can you elaborate on that please?