I have a number of mediocre quality from a decade or two back pdf's that OCR fails on. it could be that - it also for some reason occasionally fails on locked or adobe special encrypted locked pdf files. I'd check on the web end of things just to make sure this isn't a metadata download fail on the client side.