Progress in ecology depends on sound and robust methods for making inferences from data. However, typical statistical practice in the early twenty-first century frequently contradicts with what is commonly considered good scientific practice. Here we summarise an alternative statistical workflow which tries to facilitate the integration of robust statistical inference for ecologists and natural scientists in general. We are aiming to provide an overview of methods and tools which have emerged in recent years and demonstrate how these might benefit the field of ecology in particular. The presented holistic approach to statistical inference includes open science policies, reallocation of credibility instead of significance testing, averaging of information across statistical models, and the constant testing of underlying assumptions and final scientific output. We then demonstrate this approach through a worked example from our current research, where we test the dependency of turnover in pollen communities on temperature.