Files

T

Antigravity Agent 9a135bd19d docs(10_Wiki): Topic_Business/General/Graphic/Programming을 Topics/ 하위로 이동

최상위 10_Wiki/Topic_*였던 4개 카테고리 폴더를 10_Wiki/Topics/Topic_* 로 재배치.
콘텐츠 변경 없음(순수 폴더 이동) — Topics/ 하위 나머지 폴더는 이미 지난 커밋에서
전부 정리된 상태(잔존 항목은 에이전트 운영 상태 및 사용자가 보존을 요청한
업데이트0615/무제 3.canvas 뿐).

2026-07-05 00:44:01 +09:00

4.8 KiB

Raw Blame History

id, title, category, status, canonical_id, aliases, duplicate_of, source_trust_level, confidence_score, verification_status, tags, raw_sources, last_reinforced, github_commit, tech_stack

title

ARPU / ARPPU

매 한 줄

"매 ARPU 는 매 user base 전체 의 monetization 효율, 매 ARPPU 는 매 paying user 의 willingness-to-pay". 매 두 metric 의 ratio = conversion rate. 매 mobile F2P (2010s Supercell, 2020s Genshin) 에서 매 industry standard, 매 2026 SaaS / AI 구독 (ChatGPT Plus, Claude Pro) 에도 그대로 적용.

매 핵심

매 정의

ARPU = Total Revenue / Total Active Users (DAU 또는 MAU 기준).
ARPPU = Total Revenue / Paying Users.
Conversion Rate = Paying Users / Active Users = ARPU / ARPPU.

매 Time window

Daily ARPU (ARPDAU): revenue_day / DAU_day — 매 noisy, 매 7-day rolling avg 권장.
Monthly ARPU (ARPMAU): revenue_month / MAU_month — 매 industry standard.
LTV-adjusted ARPU: 매 cohort 기반 — 매 churn 반영.

매 응용

F2P 게임: 매 ARPPU $10–50, 매 conversion 1–5% → ARPU $0.10–2.50.
SaaS B2C: 매 conversion 5–15%, 매 ARPPU $5–30 → ARPU $0.50–4.
AI 구독: 매 ARPPU $20, 매 conversion 5–10%.

💻 패턴

Pattern 1: SQL ARPU/ARPPU 계산

-- BigQuery: monthly ARPU & ARPPU
WITH monthly AS (
  SELECT
    DATE_TRUNC(event_date, MONTH) AS month,
    user_id,
    SUM(revenue_usd) AS user_revenue
  FROM events
  WHERE event_date >= DATE_SUB(CURRENT_DATE(), INTERVAL 12 MONTH)
  GROUP BY month, user_id
)
SELECT
  month,
  COUNT(DISTINCT user_id) AS mau,
  COUNTIF(user_revenue > 0) AS paying_users,
  SUM(user_revenue) / COUNT(DISTINCT user_id) AS arpu,
  SAFE_DIVIDE(SUM(user_revenue), COUNTIF(user_revenue > 0)) AS arppu,
  SAFE_DIVIDE(COUNTIF(user_revenue > 0), COUNT(DISTINCT user_id)) AS conv_rate
FROM monthly
GROUP BY month
ORDER BY month;

Pattern 2: Cohort ARPU (LTV-style)

SELECT
  install_cohort,
  DATE_DIFF(event_date, install_date, DAY) AS day_n,
  SUM(revenue_usd) / COUNT(DISTINCT user_id) AS cumulative_arpu
FROM user_events
GROUP BY install_cohort, day_n
ORDER BY install_cohort, day_n;

Pattern 3: Whale segmentation

SELECT
  CASE
    WHEN user_revenue >= 1000 THEN 'whale'
    WHEN user_revenue >= 100  THEN 'dolphin'
    WHEN user_revenue >= 10   THEN 'minnow'
    ELSE 'free'
  END AS segment,
  COUNT(*) AS users,
  SUM(user_revenue) AS revenue,
  AVG(user_revenue) AS arppu_segment
FROM monthly
GROUP BY segment;

Pattern 4: Python Pareto 검증

import numpy as np
revenue = df['user_revenue'].sort_values(ascending=False).values
top_1pct = revenue[:int(len(revenue) * 0.01)].sum()
total = revenue.sum()
print(f"Top 1% contribute: {top_1pct / total:.1%}")  # 매 F2P 보통 50%+

Pattern 5: ARPDAU rolling window

SELECT
  event_date,
  AVG(revenue / dau) OVER (
    ORDER BY event_date ROWS BETWEEN 6 PRECEDING AND CURRENT ROW
  ) AS arpdau_7d
FROM daily_metrics;

매 결정 기준

상황	Use which metric
매 전체 monetization health	ARPU
매 pricing / paywall 효과	ARPPU
매 funnel optimization	Conversion = ARPU/ARPPU
매 long-term value	Cohort LTV (ARPU × retention 적분)
매 whale dependence	Top 1% revenue share

기본값: 매 ARPU + ARPPU + Conversion 매 trio 같이 보고. 매 single metric 의 misleading.

🔗 Graph

부모: 유저 평균 매출(ARPU) · 결제 사용자당 평균 매출(ARPPU)
응용: 부분 유료화(Freemium) 게임 경제 모델링 · 가상 경제 시스템
Adjacent: 이탈률(Churn Rate) · 수요와 공급(Supply and Demand)

🤖 LLM 활용

언제: 매 product analytics agent 가 매 monetization dashboard 생성 / 매 anomaly detection. 매 LLM 이 매 SQL 작성 + 매 ratio interpretation. 언제 X: 매 raw event-level data exploration — 매 BI tool (Looker, Metabase) 직접.

❌ 안티패턴

ARPU only reporting: 매 conversion 변화 의 hidden — 매 ARPPU drop + conversion rise 가 ARPU 같게 보임.
Single-day snapshot: 매 day-of-week / weekend effect → 매 7-day rolling 필수.
Mixing currencies: 매 KRW/USD/EUR mixed → 매 normalize first.
Including refunds 의 X: 매 refund 차감 안 하면 매 inflated ARPPU.

🧪 검증 / 중복

Verified (Newzoo 2025 mobile gaming report, Sensor Tower 2026 benchmarks).
신뢰도 A.

🕓 Changelog

날짜	변경
2026-05-08	Phase 1
2026-05-10	Manual cleanup — definitions + 5 SQL patterns + 매 whale segmentation

4.8 KiB Raw Blame History Unescape Escape