Retention

Day 7 retention (On or After):
User returned on Day 7, 8, 9, 10... → Retained ✓
User returned on Day 1-6 only → NOT Retained ✗

Day 7 retention (On):
User returned exactly on Day 7 → Retained ✓
User returned on Day 6 or Day 8 → NOT Retained ✗

Daily cohorts over 14 days:
├── Jan 1 cohort: Users born on Jan 1
├── Jan 2 cohort: Users born on Jan 2
├── ...
└── Jan 14 cohort: Users born on Jan 14

See: Day 1, Day 2, ... Day 14 retention for each

Weekly cohorts over 8 weeks:
├── Week 1 cohort: Users born Jan 1-7
├── Week 2 cohort: Users born Jan 8-14
├── ...
└── Week 8 cohort: Users born Feb 19-25

See: Week 1, Week 2, ... Week 8 retention for each

Monthly cohorts over 6 months:
├── January cohort: Users born in January
├── February cohort: Users born in February
├── ...
└── June cohort: Users born in June

See: Month 1, Month 2, ... Month 6 retention for each

Custom brackets: [1, 3, 7, 14, 30, 60, 90]

View retention at:
├── Day 1
├── Day 3
├── Day 7
├── Day 14
├── Day 30
├── Day 60
└── Day 90

           │ Day 0 │ Day 1 │ Day 7 │ Day 14 │ Day 30
───────────┼───────┼───────┼───────┼────────┼────────
Jan 1-7    │ 1,000 │  42%  │  28%  │   22%  │   15%
Jan 8-14   │   850 │  45%  │  30%  │   24%  │   18%
Jan 15-21  │ 1,200 │  38%  │  25%  │   20%  │    -
Jan 22-28  │   950 │  40%  │  27%  │    -   │    -
───────────┼───────┼───────┼───────┼────────┼────────
Average    │ 1,000 │  41%  │  28%  │   22%  │   17%

100% ─┬────────────────────────
      │\
      │ \
  50% ─┤  \___
      │      \_____
      │            \________
   0% ─┴─────────────────────────
      Day 0  7  14  21  28  35

100% ─┬────────────────────────
      │\
      │ \__ Pro users
      │    \_______________
  50% ─┤
      │  \
      │   \___ Free users
      │       \______
   0% ─┴─────────────────────────

Birth Event: signup_completed
  └── where: signup_source = "twitter"

Only users who signed up from Twitter

Return Event: transaction
  └── where: chain = "ethereum"

Users must transact on Ethereum to count as retained

User Filter:
├── plan = "pro"
├── AND country = "US"

Retention for US Pro users only

Retention by plan:
├── Pro: 45% → 32% → 28%
├── Free: 38% → 22% → 15%
└── Trial: 25% → 12% → 8%

Retention by signup_source:
├── Twitter: 40% → 28% → 22%
├── Discord: 52% → 38% → 32%
└── Google: 35% → 20% → 14%

Are newer cohorts retaining better?
├── Jan cohorts: 40% → 25% → 18%
├── Feb cohorts: 42% → 28% → 22%
└── Mar cohorts: 45% → 32% → 25%

→ Yes! Retention is improving

Birth Event: First transaction
Return Event: Any transaction

Question: Do users keep transacting?

Birth Event: First swap
Return Event: swap_completed
Filter by: chain = "arbitrum"

Question: Do Arbitrum users keep swapping?

Birth Event: First NFT mint
Return Event: Any NFT activity (mint, transfer, sale)

Question: Do minters become collectors?

Birth Event: used_advanced_feature
Return Event: any_active_event

Compare to:
Birth Event: did_not_use_advanced_feature
Return Event: any_active_event

Question: Does the feature improve retention?

✅ Curve flattens (stabilizes)
✅ Newer cohorts retain better
✅ Power users have much higher retention
✅ Key features correlate with retention

❌ Curve never flattens (constant churn)
❌ Newer cohorts performing worse
❌ No difference between segments
❌ Day 1 retention below 30%

✅ Good birth events:
- signup_completed (clear milestone)
- wallet_connect (clear commitment)
- first_transaction (value delivered)

❌ Poor birth events:
- page_view (too broad)
- click (too minor)

Social app: Return = any_active_event
DeFi protocol: Return = transaction
Marketplace: Return = purchase OR listing
SaaS: Return = feature_used

✅ Actionable:
- By signup_source (optimize marketing)
- By completed_onboarding (improve onboarding)
- By user_tier (price/feature adjustments)

❌ Not actionable:
- By random_id
- By timestamp

For Day 30 retention with weekly cohorts:
Need 5-6 weeks of data minimum

For quarterly analysis:
Need 6+ months of data