In the analysis of survey data, common objectives include estimating the population average, tracking time trends, and comparing population subgroups.  When samples are taken over time, we can estimate using only the present time data or also include historical data. However, when the characteristic is drifting over time and sample sizes are small, the decision to include historical data trades precision for bias. We propose regulating the bias-variance trade-off using Weighted Estimating Equations based on a suitable Generalized Linear Model that incorporates covariates. A customer loyalty survey for a smartphone vendor will be presented and resulting present time estimates of Net Promoter Score will be compared across various approaches applied to example data and simulated data.