We consider the problem of limited feedback codebook design for multi-user beamforming in 
the spatially correlated broadcast setting. In contrast to existing designs based on
isotropic directional quantization (often used in the form of random vector quantization-
based codebooks), we leverage ideas from the correlated single-user setting in this work. 
While the codebook is designed based on statistics-induced localizations, the key difference 
lies in the metric used for best codeword selection at the user end. Each user estimates its 
signal-to-interference and noise ratio (SINR) and feeds back the index of the codeword that 
maximizes this estimate. Such codebook constructions with a primary goal of interference 
management provide significant performance improvement over existing codebooks in the 
practically relevant low-feedback regime. Extensions to the interference channel setup are 
also provided.