Multiple techniques have been developed and used to characterize the width of a written track in perpendicular recording. This work investigates written tracks using the WPE (write plus erase) test based on writing the main track at a given density and trimming both sides of the track multiple times with another density. The edge of the original track, after each trimming step, is characterized by a narrow band measurement of the read-back signal as a function of cross-track position. The use of a narrow band filter improves the test SNR and helps separate the track into 2 sections: A good region (track center) and a degraded region (track edge). Test and micro-magnetic analysis were used under multiple linear densities for the main track and for the trimming tracks. The results indicate that the WPE value decreases as the density increased between 200 kfci and 1500 kfci. For densities lower than 200 kfci and higher than 1500 kfci, the WPE value does not change significantly. When the main track was fixed at 200 kfci, increasing the trimming density resulted in a higher WPE. Using a micro-magnetic model and spin-stand tests, it is demonstrated that the apparent increase in WPE is caused by a region with opposite polarity at the edge of the main track (in the media) produced by the magneto-static (demagnetization) field generated by the main track. The section with opposite polarity is found to be coherent with the main track. During the read-back process, the field from the opposite polarity region in the media is integrated by the reader (read sensitivity function) and subtracted from the main track signal. The result is a loss in signal at the track edge that yields a larger WPE value. This phenomenon could have a direct impact on shingled recording.