Table smells in SQL Server are just like code smells. They are just an indication that something may need to be checked or investigated. They are easy to check as well, because you can generally ferret through the object catalog metadata views to flush out the aspects that just don’t smell right. In this blog, I show the sort of query I’ll use. Actually, I generally use rather more strict criteria because I’d be concerned about tables that don’t seem to be making full use of constraints, and tables that don’t have ‘soft’ dependencies (aren’t referenced in code within the metadata, from views, procedures or functions). I like to have a result that displays the names of the tables that look suspect, along with the list of the ‘smells’.
This is a great start. As Phil notes in the comments, it’s not necessarily that these are wrong so much as that if you see them, there ought to be a specific reason for it to be this way.