Knowledge of life‐history parameters is frequently lacking in many species and populations, often because they are cryptic or logistically challenging to study, but also because life‐history parameters can be difficult to estimate with adequate precision. We suggest using hierarchical Bayesian analysis (HBA) to analyze variation in life‐history parameters among related species, with prior variance components representing shared taxonomy, phenotypic plasticity, and observation error. We develop such a framework to analyze U‐shaped natural mortality patterns typical of mammalian life history from a variety of sparse datasets. Using 39 datasets from seals in the family Phocidae, we analyzed 16 models with different formulations for natural morality, specifically the amount of taxonomic and data‐level variance components (subfamily, species, study, and dataset levels) included in mortality hazard parameters. The highest‐ranked model according to DIC included subfamily‐, species‐, and dataset‐level parameter variance components and resulted in typical U‐shaped hazard functions for the 11 seal species in the study. Species with little data had survival schedules shrunken to the mean. We suggest that evolutionary and population ecologists consider employing HBA to quantify variation in life‐history parameters. This approach can be useful for increasing the precision of estimates resulting from a collection of (often sparse) datasets, and for producing prior distributions for populations missing life‐history data.