Animal studies are about models and proof of concept, not direct replication. It’s silly to get hung up on that aspect.
Comparative Medicine is conducted with an awareness of the necessity for additional translational research, and while pharmaceutical outcomes are potentially total dead ends due to deeper biochemical pathway differences (which are strongly rooted all the way down to genetic differences, even across human subpopulations), one can easily get a sense of the value you of animal studies via trauma and gross anatomical studies.
A surgeon won’t make the same cuts and blood vessel clamps in a porcine heart transplant as with a human, but if you can transplant one pig’s heart into another pig, human to human is next. Curveballs like blood transfusion (types and factors) and immuno-suppression/rejection add overhead to making a transplant work in humans, but the answers to the big questions all say it will work.
So too with mice and rats and dogs and primates. To a degree behavioral realities fit too, although humans have a far deeper grasp of abstract concepts, so we’ll make leaps of cognition that render some animal models invalid.
Humans and animals are both prone to compulsive behaviors, addiction, self harm, but how we get there is often a little different since society and the rule of law can lend layers of social padding that give us a second chance to rescue ourselves before we violate norms and resort to pathological patterns. So animal psychology compares along bolder, broader strokes and lines than broken bones and wound care.
So 80% failure is just considering direct replication of practice, like a vet going and doing the same thing to a person. Sometimes the vet can save the day, but other times a human has a very different standard of care.