Cohort analysis — the retention curve a single average hides
Cohort analysis groups users by join period and tracks each group over time, revealing decay and improvement that a blended metric averages into a flat, misleading line.
Not yet scoped.
Why
The PoC builds a retention triangle from event data and shows how a rising blended number can hide worsening new-cohort retention.
How it works
Not yet built.
Where it lands in Jayverse
Rabbit: retention by signup cohort, not one blended rate. Track portal wallet activations by join-week cohort so a rising overall activation number can't hide a worsening new-cohort retention curve.
gitboard: add a retention triangle, not a single DAU/WAU tile. The service dashboard should chart each cohort's retention over time per service, since one averaged metric is exactly what this PoC warns against.
Key expressions
Words and phrases from this page worth keeping, with the Korean meaning and the sentence they come from.
Expression
뜻 · 쓰이는 자리
average into
(뭉뚱그려) 평균 내서 ~로 만들다 · 세부 편차가 평균 계산 과정에서 사라져버릴 때. "averages into a flat, misleading line"
retention triangle
리텐션 삼각형(코호트별 잔존율을 나타낸 표) · 가입 시기별 그룹의 잔존율을 시각화한 표를 가리킬 때. "builds a retention triangle from event data"
track over time
시간에 걸쳐 추적하다 · 한 그룹을 여러 시점에 걸쳐 계속 관찰할 때. "tracks each group over time"
hide (v.)
감추다, 드러나지 않게 하다 · 전체 지표가 좋아 보여도 그 뒤의 악화를 숨길 때. "can hide worsening new-cohort retention"
decay and improvement
하락과 개선 · 그룹별로 서로 다른 방향의 변화가 동시에 존재할 때. "revealing decay and improvement"