We will soon have the technological capability to measure the atmospheric composition of temperate Earth-sized planets orbiting nearby stars. Interpreting these atmospheric signals poses a new challenge to planetary science. In order to understand the degree to which we will learn of terrestrial planets from their atmospheres, I use the mystery of early Mars as a case study. I describe how we leverage models to learn of Mars’ ancient past, and make the case that this will be a very similar problem to the characterization of terrestrial planets with JWST. Then, I lay the framework for an information content-based approach to optimize our observations and maximize the retrievable information from exo-atmospheres. First I test the method on observing strategies of the well-studied, low-mean-molecular weight atmospheres of warm-Neptunes and hot Jupiters. Upon verifying the methodology, I finally address optimal observing strategies for temperate, high-mean-molecular weight atmospheres (Earths/super-Earths) and discuss what we can hope to learn from terrestrial exoplanets.