You're fragmenting the DNA at known sequences, so you sort of know where you're cutting it.
An example would be, say you've extracted some mRNA, reverse transcribed, and made cDNA and want to see if it contains a gene. IF you know that gene has 1,500 base pairs between two different restriction locations (assuming they only occur once in the plasmid/chromosome, whcih is unlikely but humor me for this example) then you can digest the gene, perform agarose gel electrophoresis with the sample against a known ladder, and see if you've got a 1,500 base pair fragment. If so, then you know you might have that gene you're looking for and may feel like having it sequenced to confirm the identity.