A client once sent a draft with forty-one footnotes and a note attached: see how thorough we were. The number was real. So, unfortunately, was the gap between three of those footnotes and the sentences they were attached to.
One cited a corporate backgrounder for the company's founding year; the backgrounder covered the company's history in a paragraph that never mentioned a year at all, the date had been added from memory and the citation placed nearby out of habit. Another cited a trade profile for the phrase “an industry leader”; the profile called the company “one of several players,” a rather different claim. The third pointed to a page that had since been edited, and the archived version, when we found it, said nothing about the subject at all.
All quotations, and any material whose verifiability has been challenged or is likely to be challenged, must include an inline citation that directly supports the material.
The word doing the work there is directly. A citation at the end of a sentence answers one question, does something exist at that link. It does not answer the question a reviewer actually asks, does that source, read in full, say this. A page can carry forty-one footnotes and still fail verification on three sentences, because density was never the standard.
So we stopped counting citations and started reading them, every one, against the exact sentence it sits beside, not the paragraph, not the general subject. A source that supports the company's size does not automatically support its founding year. A source that discusses the product does not automatically support a claim about market share. Reviewers make this same check, quietly and without announcement, and a mismatch found there tends to cost more credibility than a missing citation ever does, because it looks less like an oversight and more like a claim nobody checked.
The fix is unglamorous: read the source, then read the sentence, then ask whether a stranger with no context would agree they say the same thing. It is slower than counting footnotes. It is also the only test that was ever actually being applied.
