You have to look at opportunity cost. There is significant dev time that is spent supporting those configuration mixes that relatively few people will use.
If that dev time could be spent on other features that will be appreciated by more people, then that is a decision you can make while still adhering to the principle of "users-first".
If that dev time could be spent on other features that will be appreciated by more people, then that is a decision you can make while still adhering to the principle of "users-first".