Sometimes when you "AND," "OR," or "Exclude" attributes in your audience logic, the projected audience size may change unexpectedly.
This can be due to attribute values in your audience being asked in different surveys– or what we call survey waves. If the attribute values in your audience logic spans across different survey waves, the projected audience size shown is our best estimate. This may change given the sub-set of survey respondents used to display the audience size number.
Different audiences will use different sub-sets of data based on in which survey the attributes were asked.
In the past, if you tried to build an audience with attribute values that spanned across different survey waves - you'd get a big error! But now, due to a recent update to our methodology, we're able to estimate the size of audiences built with attribute values across different survey waves.
It can be especially true that you'll see an audience size get smaller when building audiences using TV shows. When we field a survey wave, we ask about current scenarios. For example, what TV shows do you watch? The list of TV shows will be limited to what's on air at the same time we are fielding the survey. So, as you “OR” several TV shows together, it’s very possible they will have been asked among several survey waves. As that’s based on estimation, at times it won’t result in an audience size increase.