What does XMP mean for your PDF?
XMP is Adobe’s metadata format, carried inside PDFs and also inside images. A PDF can hold the old-style document information fields, an XMP packet, or both — and they can disagree, which is how a "cleaned" file still shows an author in some readers.
Design and layout software writes more than names into XMP: tool versions, document IDs that link revisions of the same file, and sometimes the path the document was saved from.
Anything that claims to strip PDF metadata should clear both stores. That is the difference between a file that looks clean in one reader and one that is clean everywhere.
Why does a cleaned PDF still show an author?
Because it has two metadata stores and only one was cleared. The document information dictionary is the old key-value list; the XMP packet is a block of XML added later. A tool that clears one and ignores the other leaves a file that looks clean in the reader you checked and dirty in the next one.
XMP is plain text inside the file, which is why it survives operations that rewrite everything else, and why it is easy to inspect: open the PDF in a text editor and search for xpacket. You will usually find the producing application, its version, a document ID, and sometimes the folder the file was saved from.
Those document IDs are the quiet part. They link revisions of the same file, so two PDFs that look unrelated can be shown to have come from the same original.