Glossary
Confidence Interval
The range a repeated score could plausibly fall within, computed by resampling the runs behind it. Ships with every score built from more than one run; a single run gets a snapshot label instead.
Definition
A confidence interval is the range a citation score could plausibly have landed in, given the variation observed across the runs that produced it. It is computed by bootstrap resampling the underlying runs, not by assuming a bell-shaped distribution that AI-answer data does not reliably follow.
A citation rate reported without an interval looks precise and says nothing about how far it could have landed somewhere else on a different run. A 40% citation rate with a 95% interval of 10% to 70% is a very different claim from 40% with an interval of 35% to 45%, even though the headline number is identical. A single-run measurement does not get a fabricated interval, it gets the single-run snapshot label instead (see Run).
Same ruler
See how this is measured
This term is part of the same declared methodology every score on this site follows: which prompts, which engines, how many runs.
Related terms
Run
One execution of a prompt set against an engine. AI outputs are not fully deterministic, so a single run is a snapshot; several runs, or replications, form a range with a confidence interval.
Single-Run Snapshot
The honesty label for a score built from exactly one pass, not an average of repeated runs. Real data, but a photograph, not a trend line.
Want to see where your own brand stands?
One-time report, no subscription, delivered in days.