Traditional feature importance algorithms will tell us which features are most important across the entire population, but this one-size-fits-all approach doesn’t always apply to each individual customer. A factor that is an important driver for one customer may be a non-factor for another.