Understanding the processes underlying biodiversity is a real challenge, especially with the major biological crisis we are facing. To comprehend how and why living beings diversify, studying phenotypic variation among individuals is essential. Two levels of variation are mainly used to assess differences between individuals: variation in mean and in plasticity. Another component of variation has received limited attention: predictability. Such variation provides information on how predictable an individual is in expressing a trait. It was long considered to be homogeneous across individuals of a species and without biological interest. Nevertheless, variation has been measured and its evolutionary implication has been shown, mainly in livestock. Therefore, it is crucial to extend predictability studies to free-living animal and integrate this component into evolutionary research. Through the long-term data set on yellow-bellied marmots (Marmota flaviventer) and advanced analytic tools, I plan to assess the implication of predictability in evolutionary processes. I believe that being predictable or not may be part of adaptive strategies and under natural selection. Hence, it should be considered to get a more complete knowledge of evolutionary mechanisms. In a context where human pressure on wildlife is constantly increasing, it would help biologists to predict more accurately population size variations and adaptive dynamics of species.