When setting up a QF session, i was wondering which is faster, a Zone OCR in the pre-classification with token identification, or Form Identification.

I have 33 very similar looking forms, with 2 main distinct types i need to worry about. I figured I may get away with 1 or 2 document classes, but unsure what will be the best for optimization with that many samples

Basically, I have been provided 33 samples, and noticed that the first page has two possible types, and once i can separate those apart, then the only difference in the rest of the samples is the information in one of the title locations on the first page. 

If i use form identification, will I need 33 document classes? Would that make the ZoneOCR processing the better choice? 

I would appreciate anyones help in figuring out what would be more efficient as with  this many documents, i want to make sure i get it started properly before I do too much work.