In this talk, we present new, top-down analysis methods for examining and possibly verifying energy savings claims from energy efficiency (EE) policies and programs. These new methods utilize monthly utility sales data for electricity and natural gas that have been collected by the Energy Information Administration since 1990. The methods create a "statistical magnifying glass" to examine bottom-up energy savings claims using empirical energy sales data. The methods that we present are similar to an econometric analysis of panel data, but with a twist. The twist in our method is that instead of fitting our statistical models directly to raw data, the data is preprocessed with a non-linear statistical and temporal filter to produce low-noise, high-signal normalized post-processed data for the policy and program correlation analysis. The mathematical justification for the data filtering derives from the very particular temporal signature of typical EE policy and program energy savings estimates. The statistical pre-filter takes advantage of the temporal properties of energy savings claims and removes variability in the data that bears little or no relevance to long term EE policy impacts. Components of variability removed by the filtering include high frequency noise, short-term weather variability, long term population trends and inter-annual economic variability. The resulting low noise data signal is decomposed into five components for both the residential and commercial sector. The weather-normalized and noise-filtered energy components provided in the post-processed data are (1) hot-weather correlated electricity, (2) cold-weather correlated electricity, (3) weather-independent electricity, (4) cold-weather correlated natural gas and weather-independent natural gas. Given the relatively low noise levels in the post-processed data and the more detailed treatment of weather-induced energy use variability, we find a higher degree of detection sensitivity than other studies have found for EE-program-correlated energy use trend changes. We present some preliminary correlation results and discuss further plans for validation and testing of this new variety of top-down EE policy and program impact detection calculation.