Most teams misunderstand retention

Segments activate at different speeds. Averaging them into one curve is how teams ship generic fixes, then wonder why the line barely moves.

Retention without leading behaviors is guesswork. If you cannot name what people do before they stay, you are narrating a chart, not designing.

The version that works is plain: lifecycle definitions you keep, cohort discipline, and a loop where product changes tie to what you can observe.