Global patterns are key to understanding macroevolution.

Regional diversification patterns are increasingly being used to investigate spatial heterogeneity in biodiversity evolution and its drivers. However, estimating and interpreting regional diversification rates is complicated by the fact that regional diversity fluctuations reflect multiple biological processes and may be more sensitive to temporal sampling bias than global patterns. Here, we assess the biological fidelity and sampling sensitivity of regional diversification rates using fossil occurrences of marine invertebrates. We show that immigrations and extirpations account for up to 76% of apparent plate-level generic turnover. Moreover, regional diversity fluctuations are more strongly shaped by preservation, sampling, and tectono-sedimentary history than global patterns. In widely distributed groups, global analyses integrate data across regions and environments, dampen region-specific biases, and recover a higher proportion of shared biological signal. Global analyses therefore remain indispensable for inferring macroevolutionary dynamics over deep time.