Welfare is the state of an animal on a continuum, from poor to good, so many decisions about it are decisions of degree, such as how much feed, space or environmental enrichment should be provided. Other decisions are more discrete, such as whether animals should be kept in cages. However, in practice, many such decisions also involve a range of possibilities — such as whether laying hens should be kept in conventional cages, furnished cages, other housed systems or free range — so that decisions within the range are also of degree. Furthermore, in broader contexts, such as husbandry standards for farm animals, decisions are needed as to how many criteria are to be addressed, which are also decisions of degree. Similarly, decisions about which species to protect and from how early in individual development they need protection are to some extent categorical. This is sometimes referred to as 'line drawing.' However, this mainly refers to whether or not animals are sentient, and sentience is not clearly distinguished from other aspects of animals' cognition and responses, so there is no conclusive boundary between 'haves' and 'have nots.' So, these decisions are also of degree: is there sufficient evidence to 'move the line' further? When there are pressures against change, such as financial cost, should welfare advocates ask for small or large changes? The answer to this question will depend upon circumstances. But discussion of different circumstances suggests that compromise, realism, gradualism and pragmatism are all important in achieving improvements in animal welfare, while noting that other tactics also contribute in particular contexts.