If retrieving metadata works for the others, you're very likely not doing anything wrong. Zotero is just not able to find suitable metadata for those PDFs.
There can be all types of reasons -- they don't contain OCRd text, you OCRd the text yourself, so it doesn't conform to what's searchable on google scholar (which is what Zotero relies on), the article isn't archived on google scholar, Zotero picked a range of unsuitable search phrases from the article to find it.

You can look up the metadata online, import it via the Save to Zotero icon and then manually. Otherwise, yes, you'll have to do it manually. Depending on the nature of PDFs you're importing, I'd expect that to be the case for anywhere between 5 and 30% of files. Still saves you a bunch of time.

Thanks so much for responding so quickly. I suspected there would be times when I'd have to enter manual citation info, I just wish there weren't! The two files that haven't been cooperative so far are likely not archived in Google Scholar.

Kind of depends what numbers you're looking at and what's important to you. One of the key disadvantages of retrieve metadata is that you don't get abstracts and you rarely get DOIs. The former are rarely used in citations, but many people really want to have them. The latter are an increasingly common citation requirement (notably e.g. in APA style).

Otherwise google has been getting better about the data and it's pretty good, but not perfect.

edit: and the downside, of course, is that if you currently only have the PDFs, doing this in your browser is kind of laborious -- search for every article in a suitable database (don't use google scholar, ovbiously -- that won't improve over the PDF), then import.

"so much to do, so little time, so little patience"

Various work-arounds are listed in:
https://forums.zotero.org/discussion/6019/finding-unindexed-items/