While animal sentience research has flourished in the last decade, scepticism about our ability to accurately measure animal feelings has unfortunately remained fairly common. Here, we argue that evolutionary considerations about the functions of feelings will give us more reason for optimism and outline a method for how this might be achieved.