This study presents results using a novel inverse method to deduce collision-coalescence rates. Using observations of the cloud drop size distribution near the tops of stratiform clouds, we estimate the actual collision-coalescence rates found in nature. These rates do not make assumptions about which processes dominate, and instead act as a constraint on the total collection rates exhibited in real clouds as they are the values necessary to generate the observed cloud drop spectra. The inverse procedure is validated using an analytical box model that considers only gravitational sedimentation. The procedure is able to satisfactorily reproduce the collection kernels of the analytical model. Our results are reported as enhancements in collection rate relative to that expected by only gravitational sedimentation for the first approximately ten collision events. We find that observed collision-coalescence rates near stratocumulus top are at least one order of magnitude greater than differential sedimentation. While turbulence effects may account for some of this enhancement, current representations of those processes do not quantitatively predict such high values. We evaluate the sensitivity of the results to the assumptions of the procedure, and thus far find the conclusions to be robust. These findings suggest that there remain critical gaps in our quantitative understanding of collision-coalescence.