Because linking happens while writing and is never reviewed after. A page published two years ago has forty inbound links; one published last month has one, and both are supposed to rank.
Internal link analysis maps the links between your pages and shows how link support is distributed against page value.
The crawl builds the link graph. It is compared against page value and cluster mapping, so a valuable page with almost no internal support is visible, as are links pointing at the wrong target for a cluster.
A new cluster page has one inbound link. The analysis lists existing pages that mention its topic and could reasonably link to it, ordered by their own authority.
It will not add links. It shows where support is missing and which pages could supply it; editing stays with whoever owns the page.
Inbound and outbound counts, anchor text, orphan pages, links pointing at non-canonical targets, and clusters whose links point somewhere other than the mapped page.
Yes, with whether they earn anything, which decides if it matters.
Yes, including where anchors contradict the mapping.
Yes, in Loop, with the pages attached.
Fourteen days, every module, no card. Or half an hour with someone who will run it on your own records and tell you where it does not help.