<p>Robust attribution of biodiversity change to complex human drivers is crucial for mitigating biodiversity loss and achieving conservation targets under the United Nations Global Biodiversity Framework. However, the relative effects of different drivers vary dynamically across scales and contexts, requiring a targeted yet flexible causal framework that compares competing, context-specific hypotheses, incorporates counterfactual cases, and accounts for known and unknown sources of variability. In this Perspective, we explore how biodiversity change attribution could better harness existing and emerging ecological methods to overcome challenges and uncertainties in causal analysis and applications. Attribution can be accomplished either retrospectively or prospectively, using a variety of observational, experimental and process-based modelling approaches. These approaches each have strengths and limitations, and when integrated, they can offer complementary lines of evidence to increase confidence in attribution. Broader adoption of a causal, multivariate and multiscale attribution framework will better equip conservation science to guide actions on drivers and achieve biodiversity targets.</p>