Propagation effects in the interstellar medium and intrinsic profile changes can cause variability in the timing of pulsars, which limits the accuracy of fundamental science done via pulsar timing. One of the best timing pulsars, PSR J1713+0747, has gone through two ‘dip’ events in its dispersion measure (DM) time series. If these events reflect real changes in electron column density, they should lead to multiple imaging. We show that the events are well fitted by an underdense corrugated sheet model, and look for associated variability in the pulse profile using principal component analysis. We find that there are transient pulse profile variations, but they vary in concert with the DM, unlike what is expected from lensing due to a corrugated sheet. The change is consistent in shape across profiles from both the Green Bank and Arecibo radio observatories, and its amplitude appears to be achromatic across the 820-MHz, 1.4-GHz, and 2.3-GHz bands, again unlike expected from interference between lensed images. This result is puzzling. We note that some of the predicted lensing effects would need higher time and frequency resolution data than used in this analysis. Future events appear likely, and storing baseband data or keeping multiple time–frequency resolutions will allow more in-depth study of propagation effects and hence improvements to pulsar timing accuracy.