I had a similar problem recently, and here's what I did:
SELECT
columnname as 'Original_Value',
LEN(LTRIM(columnname)) as 'Orig_Val_Char_Count',
N'['+columnname+']' as 'UnicodeStr_Value',
LEN(N'['+columnname+']')-2 as 'True_Char_Count'
FROM mytable
The first two columns look at the original value and count the characters (minus leading/trailing spaces).
I needed to compare that with the true count of characters, which is why I used the second LEN function. It sets the column value to a string, forces that string to Unicode, and then counts the characters.
By using the brackets, you ensure that any leading or trailing spaces are also counted as characters; of course, you don't want to count the brackets themselves, so you subtract 2 at the end.
TEXT
is deprecated and shouldn't be used. – Duhonnvarchar
column has to be, at the end of the day, composed of sequences of unicode characters, stored as UTF-16, and your method counts how many of those characters there are. – Coulometer