In modern manufacturing systems, Information Technologies (IT) are closely linked to production operational functioning. To be a reliable service provider, IT Departments must ensure an adequately sized IT infrastructure, to support present and future industrial activity. As such, quantitative tools are useful for IT managers to anticipate future use of IT resources, on the basis of production-related explanatory variables. This paper proposes a modelling process to build such a tool. This modelling process is organized through two main sequences, exploratory and explanatory, and mixes qualitative and quantitative analyses. Taking advantage of the masses of data stored in modern manufacturing systems, it allows a progressive understanding of the relationships between the different domains of the IT architecture. To illustrate the application of the modelling process, this endeavour was conducted in a partnership with STMicroelectronics, to base the research upon and provide a case study. The latter established quantified links among the activities of the STMicroelectronics Manufacturing Execution System server and several manufacturing related explanatory variables. This paper demonstrates that using such statistical approaches through our modelling process, can overcome a typical issue of modern IT systems capacity planning works: establishing direct and quantified links between business and IT activities.