Deriving constraints from an N-point correlation hierarchy has certain practical advantages over field-level approaches or the use of ad-hoc higher-order statistics when analysing real data. Furthermore, the use of (N>3)-point correlators for projected LSS tracers has lagged behind similar analyses in spectroscopic surveys and CMB experiments. In this talk I will present a simple, fast, and accurate estimator of the projected bispectrum, and the results of its application to existing projected clustering and CMB lensing data, as a way to quantify the potential gains for cosmological analyses with near-future datasets.