Measurement is a key aspect for Information Systems assessment in all its development phases. Different artefacts allow the application of different kinds of metrics. For instance, when the software system is already developed, test metrics such as benchmarks can be used. However, in the initial phases, metrics can only be applied over model artefacts and, then, structural metrics are especially useful because they allow measuring different properties by taking into account the structural elements of the model. In order to explore how these metrics can be defined and reused, in this paper we analyse several structural metrics over different kinds of models and we observe that they share similar characteristics. Based on them, we establish a meta-modelling approach, which includes several guidelines for the definition of the structural metrics and their reuse. In order to exemplify the approach we define and reuse two different structural metrics over the i* framework.