Yeah, as control-z said, there are several ways of storing scans as PDFs. You can save it as just the plain image, or have the text OCRed and save only the text, which is obviously a lot smaller. Or you can do both, which recognizes the text and then saves it under the image, so you can still search and select text, but you don’t have to worry about OCR errors as much because what you see is the original image. If it used to save the whole image but now OCRs it and saves only the text, that would be a lot smaller. The worst (in my opinion) is when a lot of the text is OCRed and saved as only text, and looks like a normal document, but every so often there are words that are the original scanned image, which usually means there’s a sudden different background color, different font, and different size, just for that one word. Very annoying.
Or, you know, if it used to be set to scan at a higher resolution, that’s another reason it might be smaller now. What software do you use to do the actual scan?