Drop-off
Drop-off is the point at which users abandon a flow, page, or session, measured as the share who leave at each step rather than continuing. Every funnel has drop-off; the analytical question is whether a given step's rate is anomalous against its baseline, its history, or its peers.
How it works
Drop-off is computed per boundary: of users who reached step N, the fraction that never reached N plus 1 within the analysis window. Interpretation requires context that aggregate rates omit, which is where session data earns its keep: drop-off sessions can be segmented (see session segmentation) by device, error presence, or frustration signals, and the sub-segments compared. A step whose drop-off concentrates in sessions containing error clicks has a bug; one whose drop-off is uniform has a design or intent problem. The distinction changes who fixes it.
Where you encounter it
Funnel reviews, release monitoring (a drop-off spike after a deploy is a regression siren), and form analytics.