People are creatures of habit. We often consume the same item again, for example the same song or the same pizza. The next basket recommendation task (“predictive shopping”), this is a well-known phenomenon. How well do today’s neural recommendation methods work at deciding whether to offer a new item or suggesting something that you’ve had before? In this talk, I will report on some recent experimental findings that show that they still have a long way to go. The talk is based on joint work with Ming Li, Sami Jullien, and Mozhdeh Ariannezhad.