We present a new ${it{gating}}$ method to remove non-Gaussian noise transients in gravitational wave data. The method does not rely on any a-priori knowledge on the amplitude or duration of the transient events. In light of the character of the newly released LIGO O3a data, glitch-identification is particularly relevant for searches using this data. Our method preserves more data than previously achieved, while obtaining the same, if not higher, noise reduction. We achieve a $approx$ 2-fold reduction in zeroed-out data with respect to the gates released by LIGO on the O3a data. We describe the method and characterise its performance. While developed in the context of searches for continuous signals, this method can be used to prepare gravitational wave data for any search. As the cadence of compact binary inspiral detections increases and the lower noise level of the instruments unveils new glitches, excising disturbances effectively, precisely, and in a timely manner, becomes more important. Our method does this. We release the source code associated with this new technique and the gates for the newly released O3 data.