This paper proposes a reduced physical criterion for the onset of transonic buffet. Transonic buffet is commonly understood as a low-frequency, self-sustained instability involving shock-wave motion, turbulent boundary-layer interaction, and intermittent separation. The present work does not attempt to replace established compressible-flow theory, global stability analysis, or time-accurate simulation. Instead, it offers a compact threshold framework intended to organise known variables into a falsifiable engineering model. The central idea is that buffet onset can be viewed as a propagation-limited separation instability. In this interpretation, steady attached transonic flow must do two things at once: it must carry momentum through a shock-induced adverse pressure gradient, and it must communicate corrective pressure information through a compressible medium at finite speed. Buffet is hypothesised to begin when local disturbance generation and shock loading exceed the combined capacity of the flow to maintain pressure communication and near-wall support. The proposed onset parameter combines three elements: Mach number, a normalised adverse-pressure-gradient measure, and a boundary-layer support term based on local skin friction. The paper also separates buffet onset from post-onset oscillation frequency, arguing that the local scale governing the onset threshold is not necessarily the same as the larger global scale governing the low-frequency buffet cycle. The result is a concise, testable framework for comparison against published aerofoil and wing buffet datasets. The model is presented explicitly as a hypothesis to be challenged against data, not as a replacement for existing aerodynamic methods.