Enterprise Information Systems (IS) and how to lead Big Data projects.

Corporate IT has undergone a number of changes over the last half-century. Initially dedicated to the automation of a few tasks (the first three were inventory management, payroll and accounting), it has integrated in some firms, perhaps since the early 1990s, around the term "organization", an overhaul of processes and operations. Some fifteen years later, the prevalence of software packages, which offer tools as much as processes, has made Information Systems Departments much less active on this subject, and on the resulting innovation.

Nowadays, there are other types of project in addition to automation (transferring costs from people to technology), and when it comes to transformation or information projects, Big Data represents a value relay provided by IT Departments. As opposed to building information systems on the basis of models of the world (processing models, data models, etc.), the Big Data approach uses mathematics to find models in the data, models which, under certain conditions, enable us to get a head start on the present, and sometimes on the future, in order to improve decisions and operations.

We note that while finding models, for example Machine Learning models, that perform better than the existing ones (a mix of vision, experience and... models) is relatively easy, inserting these models into the right decision mechanisms for performance improvement is less straightforward. Following on from this observation, we present a general approach and specific methods aimed at ensuring that Big Data projects are aligned with performance improvement objectives.