We have all become accustomed to the reliability of the little blue dot on our smartphone maps, and the fusion of satellite, cellular and Wi-Fi positioning systems mean that it is rare to find ourselves stuck without a position fix. But what started life as a casual navigation tool for vehicular navigation (benefiting from the assumption that snap-to-the-road algorithms correct the positioning errors) is now being squeezed for more and more performance. Your smartphone will soon confirm which supermarket aisle you are in, and will take you right to your seat at football matches and concerts. That level of accuracy requires a very careful appreciation of what is really going on behind the little blue dot, and mere assumptions have to be left at the door. In this talk we will look at the importance of confidence estimation in the design, implementation and delivery of indoor location systems and its role in providing improved performance, not just better error estimates, for the user. We will look at how system modelling and computational grunt can provide higher positioning performance than measurements alone can, and why app developers can only take their tools so far – the true capabilities lie with the chipset providers.