Key driver analysis ranks your questions by how much they move together with the outcome you're tracking — NPS, a rating, anything on a scale — so you stop guessing which fix matters most.
A crosstab shows you how one question relates to your outcome. A real survey asks twenty. Key driver analysis looks at every eligible question at once and ranks which ones actually move together with the score you're tracking — instead of you guessing which pair to check next.
Every eligible question gets a share of the explained variance, ranked highest to lowest — so you know which few things matter most, not just the one you happened to crosstab. Part of Growth plans and above, alongside crosstab analysis.
Survey answers are almost always correlated with each other — people who rate support highly tend to rate delivery highly too. The method underneath (Johnson's relative weights) is chosen specifically because it holds up under that, where a plain regression coefficient would mislead.
Too little data, two questions answered almost identically, or numbers that can't support a reliable answer — SurveyRock says so instead of returning a ranking it can't stand behind.
Low sample size on a driver, incomplete overlap between questions, or a driver where nearly everyone gave the same answer — flagged directly on the result, not left for you to discover later.
Raise an Action directly from a driver result — give it an owner and a due date without leaving the page, and see whether the outcome moves once it's done.
Rating, NPS, slider statements, and single-answer scale questions are all eligible drivers.
Rating, NPS, or a slider statement — whatever score you're trying to move.
See which questions move together with it most, and exactly which caveats apply — low sample size, low overlap between questions, or a driver almost everyone answered the same way.