Motif enrichment analysis is checking if there are any over represented sequences through PWMs in the regions/peaks of your interest.

Foot-printing based on deep sequenced ATAC/DNAse data tells you if something is really bound to peaks of interest.

If you imagine a transcription factor is bound to some regions of DNA (peaks), when you do ATAC or DNAase experiment, as there is some bound TF, it will be difficult for the transposase or DNAase to cleave the protein bound regions of DNA which leaves "footprints" which can later be identified from the sequencing data as you would see a dip in those regions.

This is a random figure I pulled from google, which shows that there is a "dip" in DNAse signals when there is CTCF binding ( from ChIP-Seq data) suggesting that by using deep sequenced ATAC or DNAse, you can infer "true" binding events, not just presence/absence of sequence motifs which doesn't tell you if those TFs are really bound to DNA.

PS: You may need millions of reads (probably >500 million from single sample) to accurately infer "true" binding events and it works for TFs/proteins that have long residing times. For example, most of the times it works for CTCF.

A good review here.