Building reservoir models that consistently honors static and dynamic data is a difficult, if not impossible, task using traditional approaches resulting from limitations of existing tools and best practice workflows. The crux of this task has traditionally been to utilize dynamic data in the facies modelling process, which is often the cornerstone of the reservoir modelling workflow. Hence, failing to integrate the static and dynamic data measurements in the facies modelling process consistently can dramatically reduce the predictability of the generated reservoir models. In this paper, we efficiently solved this problem using an ensemble-based approach in combination with an adaptive pluri-Gaussian facies modelling scheme. We demonstrate the procedure on a medium size field with 15 years of production history. During the dynamic data conditioning, clear trends are established in the facies model throughout the reservoir, which provide a good indication of the expected facies distribution and associated connectivity. Having a thorough description and understanding of the subsurface uncertainties - especially when it comes to the facies model description - is key to improved reservoir management decisions when considering both optimal drainage strategies and well placement.