Dark matter models have different amounts and distributions of it for each galaxy we care to look at, so there is no predictive power, similar to how they had to add new epicycles to explain the motion of every new planet they observed.
Just today I heard a talk about people looking at stellar motions in ultra faint dwarf galaxies, because standard dark-model theories predict they should have centrally cusped dark-matter density profiles -- and fuzzy dark-matter models predict different profiles, which they could potentially discriminate between.
However, it is very much true that different galaxies have different amounts and distributions of dark matter. There are many broad categories, and certain models about galaxy formation try to predict these, but not all galaxies fit these models. This has two effects: for one, for any galaxy with unexpected dynamics, you can fit a dark matter distribution to explain the dynamics (but! This is not unbounded, it still has to match certain other observations). The other effect is actually in favor of a dark matter approach: it means modifying parameters of our theories in a general way has little chance to reproduce the same properties, as each different galaxy is different.