Over the last couple of years, multiple image-based localization approaches have been proposed that enable us to handle large-scale scenes both efficiently and effectively by being able to localize query images over a large range of viewing conditions in very little time. Very recently, we have shown that we can leverage these methods to make large-scale, real-time localization feasible on mobile devices such as tablets and smartphones. This recent success naturally begs the questions of whether we have already solved the image-based localization problem. In this talk, we will argue that this is actually not the case. Starting with a detailed overview over our current localization pipeline, we will analyze the shortcomings of the current state-of-the-art in image-based localization in terms of computational costs, scalability, robustness, and their ability to handle certain scene elements.