Poultry digital twins must earn their authority: Biological validity, welfare value, and responsible adoption.

Digital twins could connect poultry observations with decisions about welfare, physiology, production and product quality, but their usefulness depends on what is represented, how the representation is updated, and what decisions it improves. This critical narrative review examines poultry studies and related methodological literature to distinguish sensing performance, biological validity and decision value. I define a poultry digital twin as a persistent, dynamically updated representation of a specified physical entity or process, with a model of change and an explicit decision and feedback pathway. Seven author-defined categories organize the discussion: HatchTwin, HenTwin, CluckTwin, EggTwin, FlockTwin Population and BroilerTwin cover overlapping application domains, while ChickenTwin provides a comparative umbrella. These names do not imply seven validated systems. The evidence map separates biological reference studies and sensing components from integrated implementations, and distinguishes commercial testing from demonstrated intervention benefit. Context-dependent signals, incomplete identity, correlated measurements and limited external evaluation constrain the interpretation of welfare and production states. Building on these findings, I define staged evaluation through biological validation, prospective silent operation and appropriately designed decision trials. Outcomes should include animal-based welfare, production, workload, costs and unintended harm. Canadian examples illustrate adoption and governance under a defined jurisdiction. The proposed framework is intended to guide testable claims and proportionate use; its effectiveness as a whole remains to be evaluated.