"duplication is far cheaper than the wrong abstraction" applies just as much to teams and process as it does to code.

But also duplication is far more expensive than the right abstraction, and that is a far more common problem

If you see a process problem, solve it in one place first. Then solve it, separately, in another place.

Once you’ve solved it a few times, you can start to factor out commonalities. Early optimisation will probably just help you do the wrong thing faster.