Results from the point source contamination estimator of Eq. (5). The best-fit cross-spectra associated with shot noise are plotted in black. We note that the signal-to-noise ratio at 545 and 857 GHz is particularly low. The grey line is a prediction for the bias from the CMB lensing-infrared correlation, and has been subtracted from the spectra (plotted as black points). We see that with the subtraction of the bias from CMB lensing, the measured bispectrum-related spectrum is generally consistent either with zero or with the shape expected for shot noise.