Sometimes SIFT is limited when the content of the image and the features are so complex and have so many details.

for example, when a human face is showing alone in the image and looking directly at the camera, it will be useful to use SIFT (but still not best nowadays), but when there is a crowd, and so many faces in the image to extract key points for, SIFT will not be doing that good and it is better to use some other technique.