Any tips on how to eliminate key lookup

As mentioned by @SQLTrooper, that would be the standard way to remove a key lookup. It should be mentioned that key lookups aren't always bad, but in your case clearly one of them seem that way. What you may find is that when (assuming it was possible) you added this additional column to the index you'd find that index would appear the performance bottle neck, after all it all has to add up to 100% somehow.

There are always many ways to skin a SQL CAT 🙂 for key lookups you could take the standard approach or consider an index on that column alone, be that XML/Full Text/Non Clustered. The data could then be read from that index and joined back, potentially better than the keylookup.

It may well be worth considering the data stored in the column and trying a Full Text index – which will work with the current datatype. That said you should obviously, due to the fact it's deprecated, change the data type at some point.

My tests show that you can create a clustered index on the column, but I don't think that would be much use in this case, interesting though. What's more interesting is that after creating a schema bound view, a clustered index can't be added to that view specifically on the ntext column.

Further reading and info…

This link shows info about include columns:

http://technet.microsoft.com/en-us/library/ms190806(v=sql.105).aspx

Specifically it states that the data type you mention can't be included in an index include column, my tests show it can't be used as a key column or and include column. Paul Randall (http://blogs.msdn.com/b/sqlserverstorageengine/archive/2007/03/14/can-you-include-columns-of-type-text-ntext-or-image.aspx) states the old data types are stored in a similar way to the new data types but "the internals of how they are processed by the server are different so they [the new types] behave like regular data types".

Hope that helps!