Why Unsourced Stats Are Round Numbers (2,473-Claim Analysis)
The last digit of a percentage tells you whether anyone counted it.
How content teams keep published data accurate.
The last digit of a percentage tells you whether anyone counted it.
We scanned 5,034 data claims across 46 domains and found that only 11.8% carry a date you can actually check.
1,006 citations. 516 domains. The weight behaves like 104.
One URL, its own history as the control, and a verdict that can admit it doesn't know.
No response count gives a self-selected reader poll a margin of error, so publication comes down to the sentence you're willing to attach to the votes you already have.
Every 'most' in your archive marks a number you never measured, and the people who could supply it are already reading the sentence.
Six dates, two kinds of witness, and the one record a publisher can never rewrite.
Your archive has a history, dependencies, and production traffic. Nothing is watching the build.
Every audit tool re-checks what lives on your own page: links, meta, copy. The column that rots is the cited statistic, and it rots at its source, on a page you don't control.
Three snippets render the same chart, and only one leaves the number on your page as text a crawler can read.