Bacteria often attach to surfaces and grow densely-packed communities called biofilms. As biofilms grow, they expand across the surface, increasing their surface area and access to nutrients. Thus, the overall growth rate of a biofilm is directly dependent on its "range expansion" rate. One factor that limits the range expansion rate is vertical growth; at the biofilm edge there is a direct trade-off between horizontal and vertical growth-the more a biofilm grows up, the less it can grow out. Thus, the balance of horizontal and vertical growth impacts the range expansion rate and, crucially, the overall biofilm growth rate. However, the biophysical connection between horizontal and vertical growth remains poorly understood, due in large part to difficulty in resolving biofilm shape with sufficient spatial and temporal resolution from small length scales to macroscopic sizes. Here, we experimentally show that the horizontal expansion rate of bacterial colonies is controlled by the contact angle at the biofilm edge. Using white light interferometry, we measure the three-dimensional surface morphology of growing colonies, and find that small colonies are surprisingly well-described as spherical caps. At later times, nutrient diffusion and uptake prevent the tall colony center from growing exponentially. However, the colony edge always has a region short enough to grow exponentially; the size and shape of this region, characterized by its contact angle, along with cellular doubling time, determines the range expansion rate. We found that the geometry of the exponentially growing biofilm edge is well-described as a spherical-cap-napkin-ring, i.e., a spherical cap with a cylindrical hole in its center (where the biofilm is too tall to grow exponentially). We derive an exact expression for the spherical-cap-napkin-ring-based range expansion rate; further, to first order, the expansion rate only depends on the colony contact angle, the thickness of the exponentially growing region, and the cellular doubling time. We experimentally validate both of these expressions. In line with our theoretical predictions, we find that biofilms with long cellular doubling times and small contact angles do in fact grow faster than biofilms with short cellular doubling times and large contact angles. Accordingly, sensitivity analysis shows that biofilm growth rates are more sensitive to their contact angles than to their cellular growth rates. Thus, to understand the fitness of a growing biofilm, one must account for its shape, not just its cellular doubling time.