How often we decline to answer
Every tool here can refuse. When the filings do not contain what a number needs, we print nothing rather than a figure we cannot stand behind. This page counts how often that happens, because a claim about our own honesty is worth exactly as much as the measurement behind it.
These surfaces are measured but sit under 30 attempts in the window. A percentage from a handful of observations reads as far more certain than it is, which is the same error this page exists to avoid making, so the counts are shown and the rate is not.
The denominator is attempted derivations, not page views. Attempts and refusals are counted by the same code in the same request, so neither can drift from the other. A page served from cache never runs that code, which is precisely why traffic would be the wrong denominator: it would undercount refusals and flatter us.
Counting began on , and covers the surfaces instrumented so far rather than every tool on the site. Refusals that happen deeper than the instrumented layer are not yet counted, so the real rate is likely higher than what is shown here, never lower.
Every refusal is also filed under its ticker: what was asked, why we stopped, and the filing item that would change the answer.
Free to use. No account, no card.