When I export my PDF with images of text... the doc file just has images of text... isnt Optical Character Recognition supposed to convert this?

OCR is not part of the Document Cloud PDF services. Even in Acrobat you just explicitly run OCR it does not happen automatically when exporting an image. I believe a recent version of Acrobat did do automatic OCR on opening image files but I turned off that feature as it can ruin as many PDFs as it works successfully on. I only use OCR when I need it.