The gap between large-scale data production rate and the rate of generation of data-driven scientific insights has led to an analytical bottleneck in scientific domains like climate, biology, and so on. This is primarily due to the lack of innovative analytical tools that can help scientists efficiently analyze and explore alternative hypotheses about the data and communicate their findings effectively to a broad audience. In this article, by reflecting on a set of successful collaborative research efforts between with a group of climate scientists and visualization researchers, the authors introspect how interactive visualization can help reduce the analytical bottleneck for domain scientists.