Manipulation of host phenotype by parasites can require a collective effort from many individuals. The cost of manipulation may only be paid by the individuals actually inducing the manipulation, while its benefits are reaped by all. Here, we determine if there is genetic variation in manipulative effort among different clonal lineages of the trematode Curtuteria australis, and whether the decision to manipulate is context‐dependent. C. australis impairs the burrowing efficiency of its second intermediate host, the cockle Austrovenus stutchburyi, by encysting at the tip of the cockle's foot, which facilitates the parasite's trophic transmission to shorebirds. However, manipulative individuals at the tip of the foot are vulnerable to non‐host predators (foot‐cropping fish); in contrast, those encysted at the base of the foot, although they do not contribute to manipulation, are safe from foot‐croppers and can benefit from altered host phenotype. In an experimental study, different clonal lineages showed no significant variation in their tendency to encyst in the tip versus the base of the foot, with only the former contributing to host manipulation. However, the decision to manipulate was intensity‐dependent: the greater the number of parasites already committed to manipulation (i.e. already encysted in the foot tip), the more likely newly arriving parasites were to join them. These findings indicate considerable intraspecific variation in the strategies adopted by ‘manipulator’ parasites, with external influences determining what a parasite actually does.