What Actually Counts as a "Citation-Worthy" Claim?

If you've ever had a supervisor circle a sentence in your draft and write "citation?" in the margin, you've probably wondered where exactly the line is. Some claims obviously need a source: a statistic, a direct quote, a finding that isn't yours. Others are murkier: is "climate change is accelerating" something you need to cite, or is it common enough knowledge to state plainly?

There's no single rule that covers every field, but a few questions get you most of the way there.

Would a reader reasonably ask "says who?"

This is the simplest test. If someone reading your sentence cold, without your background in the topic, would naturally wonder where that came from, it needs a source. Numbers almost always trigger this ("global temperatures have risen 1.1°C since pre-industrial times"). So do causal claims ("X leads to Y"), comparisons ("more effective than"), and anything describing a specific study's outcome.

Claims that describe your own reasoning or your own data don't need external citation. They need you to show your work instead, which is a different problem.

Is it contestable?

Facts that are genuinely settled and universal, like water boiling at 100°C at sea level, don't usually need a citation in most contexts, because nobody credible disputes them. But the moment a claim is specific to a population, a time period, a methodology, or a school of thought, it becomes contestable, and contestable claims need backing. "Remote work increases productivity" sounds general, but it's actually a claim about a specific body of studies with real disagreement in the literature, and that needs a source.

Does the specificity outrun common knowledge?

"The brain uses a lot of energy" is common knowledge. "The brain accounts for roughly 20% of resting metabolic energy expenditure despite being about 2% of body mass" is a specific figure that came from somewhere, and your reader deserves to know where. As a rule of thumb: the more precise the number or the narrower the claim, the more it needs a citation, even if the general idea behind it is well known.

What this looks like in practice

A useful habit is to read back through a paragraph and mentally tag each sentence as either "background knowledge," "my own argument," or "someone else's finding." Anything landing in the third bucket needs a source attached before you submit. It's tedious to do by hand across a long draft, which is honestly the entire reason we built Research Verifier. It reads through your text, picks out the sentences that land in that third bucket, and finds papers that support them, so you're not doing this sentence-by-sentence audit manually.

Either way, manually or otherwise, getting in the habit of asking "would someone reasonably ask says who here?" will catch most of the gaps before a reviewer does.