Stars are regularly observed in the visible channels of the GOES Imagers for real-time navigation operations. However, we have been also using star observations off-line to deduce the rate of degradation of the responsivity of the visible channels. We estimate degradation rates from the time series of the intensities of the Imagers' output signals, available in the GOES Orbit and Attitude Tracking System (OATS). We begin by showing our latest results in monitoring the responsivities of the visible channels on GOES-8, GOES-10 and GOES-12. Unfortunately, the OATS computes the intensities of the star signals with approximations suitable for navigation, not for estimating accurate signal strengths, and thus we had to develop objective criteria for screening out unsuitable data. With several layers of screening, our most recent trending method yields smoother time series of star signals, but the time series are supported by a smaller pool of stars. With the goal of simplifying the task of data selection and to retrieve stars that have been rejected in the screening, we tested a technique that accessed the raw star measurements before they were processed by the OATS. We developed formulations that produced star signals in a manner more suitable for monitoring the conditions of the visible channels. We present specifics of this process together with sample results. We discuss improvements in the quality of the time series that allow for more reliable inferences on the characteristics of the visible channels.