Everything changes eventually - even good things must come to an end. But when looking back, it is rarely straightforward to pin down the exact timing and nature of change. When did a recession begin? When did a native animal population start to decline? When did a processing plant start operating differently?
At the time it occurs, change is often lost amongst the normal variation and it is only inretrospect that it is evident that a change has taken place. Drawing the line between expected variation and significant difference (that is, deciding whether a change has actually occurred) is a statistical challenge.