Here is a sketch of a possible consultation in the consulting unit of my department, within a faculty of science. A researcher/PhD-student (or a couple of them) from a biology/geology/etc. department has contacted us, asking for help with the analysis of data from a study. Even if the client has provided some information before the meeting, we start by asking for a description from scratch of the background, the scientific aims, the experimental/observational setup, and to be shown some data from the study. Before we can do any statistical modeling or give any sound advice about statistical methods, we must understand the client’s data and the client’s scientific questions. The client often presents the aims of the study in a general, vague form that needs specification and statistical reformulation in quantifiable units. What is really the client’s problem? And what kind of questions can possibly be answered from the type of data available? Often, clients are stimulated to think about their problems in fresh ways. We will also ask a lot of questions in order to make clear how the data were collected. What populations do the data represent? Was there a use of randomization, stratification, censoring, etc.? On what parts of the data should the focus be? Explore the data! What might be the structure of these data? (It is always fun to see a structure in data that the client has not noticed). This discussion can lead up to a tentative statistical model, with interpretable parameters, at a later stage to parameter estimation procedures and hypothesis tests, etc.