Empirical duration - doesn't this double count spread duration?

I get the concept of empirical duration but doesn’t it end up double counting spread duration?
Would you ever use empirical duration x delta benchmark rates and also spread duration x change in spread in the same calculation?

No.

No.

You could use modified (or effective) duration × ∆benchmark rate + spread duration × ∆spread.

Maybe another way to reframe the question:
Does empirical duration replace the need to also look at spread duration?
In other words, is empirical duration a drop-in replacement for modified (or effective) duration?

Would you ever do: empirical duration × ∆benchmark rate + spread duration × ∆spread ?

Yes.