The fast evolution of the QGP makes its interaction with jets an inherently time-dependent process. However, this crucial dimension is missing from traditional jet quenching measurements, which hence provide a mere average quantification of the medium properties. Jet substructure observables provide an effective strategy to access the QGP time structure. By identifying the recursive steps of a novel jet clustering algorithm (the $\tau$ algorithm) with the sequence of branchings of the parton shower, it is possible to obtain an adequate proxy for a time axis within the medium. In this work we apply this technique to label jets according to their formation time and opening angle, defining populations with enhanced sensitivity to quenching effects. In doing so, we show how the formation time-based sampling of jets minimizes the bias of their initial transverse momentum spectrum.