Just-in-Time Data Structures are an attempt to vulgarise the idea that changing the representation (i.e., implementation) of a data structure at runtime can improve the performance of a program compared to its counter part that relies on a single representation. In previous work, we developed a language to develop such Just-in-Time Data Structures. To express ``when'' to change between representations, a dedicated language construct was introduced: the swap rule. A swap rule analyses the state and usage of a just-in-time data structure and reacts as defined by a developer. Opposed to what the name suggest, swap rules are currently implemented as imperative statements woven into the codebase. Their intend, however, is declarative and therefore we think that swap rules should become real declarative rules. This extended abstract presents Just-in-Time Data Structures as a case for applying state-of-the-art in low overhead dynamic analysis. Changing from an imperative to a declarative implementation of swap rules will allow for more efficient execution of our programs by reducing the overhead of continuous analysis.