dubs-server 1.0.0

package/scripts/analyze-cohort-data-heroku.js

@@ -0,0 +1,276 @@
+/**
+ * Cohort Retention Data Analysis - Heroku Production
+ * 
+ * This script analyzes the production Heroku database.
+ * 
+ * Usage:
+ *   heroku config:get DATABASE_URL --app dubs-server-prod > /tmp/db_url.txt
+ *   DATABASE_URL=$(cat /tmp/db_url.txt) node scripts/analyze-cohort-data-heroku.js
+ * 
+ * Or use the Heroku CLI directly:
+ *   heroku run node scripts/analyze-cohort-data-heroku.js --app dubs-server-prod
+ */
+
+const { Pool } = require('pg');
+
+async function analyzeCohortData() {
+  const DATABASE_URL = process.env.DATABASE_URL;
+  
+  if (!DATABASE_URL) {
+    console.error('❌ DATABASE_URL environment variable not set!');
+    console.error('   Set it with: export DATABASE_URL=<your-heroku-postgres-url>');
+    process.exit(1);
+  }
+  
+  const isProduction = DATABASE_URL.includes('amazonaws') || DATABASE_URL.includes('heroku');
+  
+  const pool = new Pool({
+    connectionString: DATABASE_URL,
+    ssl: isProduction ? { rejectUnauthorized: false } : false,
+    max: 2, // Use minimal connections for analysis
+  });
+
+  console.log('\n🔍 COHORT RETENTION DATA ANALYSIS (PRODUCTION)\n');
+  console.log('=' .repeat(80));
+  console.log(`Database: ${DATABASE_URL.split('@')[1]?.split('/')[0] || 'unknown'}`);
+  console.log('=' .repeat(80));
+  
+  try {
+    // 1. Check for registration events
+    console.log('\n📊 1. REGISTRATION EVENTS');
+    console.log('-'.repeat(80));
+    
+    const registrationEvents = await pool.query(`
+      SELECT 
+        log_type,
+        COUNT(*) as total_events,
+        COUNT(DISTINCT user_id) as unique_users,
+        MIN(created_at) as first_event,
+        MAX(created_at) as last_event
+      FROM audit_logs
+      WHERE log_type IN ('registration_started', 'registration_completed', 'registration_submitted')
+      GROUP BY log_type
+      ORDER BY total_events DESC
+    `);
+    
+    if (registrationEvents.rows.length === 0) {
+      console.log('❌ No registration events found in database!');
+      console.log('   Cannot build cohort analysis without registration data.');
+      await pool.end();
+      return;
+    }
+    
+    console.log('\nRegistration Events Summary:');
+    registrationEvents.rows.forEach(row => {
+      console.log(`  ${row.log_type}:`);
+      console.log(`    - Total Events: ${row.total_events}`);
+      console.log(`    - Unique Users: ${row.unique_users}`);
+      console.log(`    - Date Range: ${row.first_event?.toISOString().split('T')[0]} to ${row.last_event?.toISOString().split('T')[0]}`);
+    });
+
+    // 2. Check date range and volume
+    console.log('\n\n📅 2. SIGNUP DATE DISTRIBUTION (Last 30 days)');
+    console.log('-'.repeat(80));
+    
+    const dailySignups = await pool.query(`
+      SELECT 
+        DATE(created_at) as signup_date,
+        COUNT(*) as signups,
+        COUNT(DISTINCT user_id) as unique_users
+      FROM audit_logs
+      WHERE log_type = 'registration_completed'
+        AND user_id IS NOT NULL
+      GROUP BY DATE(created_at)
+      ORDER BY signup_date DESC
+      LIMIT 30
+    `);
+    
+    if (dailySignups.rows.length === 0) {
+      console.log('  No signup data found!');
+    } else {
+      dailySignups.rows.forEach(row => {
+        console.log(`  ${row.signup_date.toISOString().split('T')[0]}: ${row.signups} signups (${row.unique_users} unique users)`);
+      });
+    }
+
+    // 3. Weekly cohort summary
+    console.log('\n\n📊 3. WEEKLY COHORT SUMMARY');
+    console.log('-'.repeat(80));
+    
+    const weeklyCohorts = await pool.query(`
+      SELECT 
+        DATE_TRUNC('week', DATE(created_at)) as cohort_week,
+        COUNT(*) as signups,
+        COUNT(DISTINCT user_id) as unique_users
+      FROM audit_logs
+      WHERE log_type = 'registration_completed'
+        AND user_id IS NOT NULL
+      GROUP BY DATE_TRUNC('week', DATE(created_at))
+      ORDER BY cohort_week DESC
+      LIMIT 12
+    `);
+    
+    console.log('\nLast 12 weeks of signups:');
+    if (weeklyCohorts.rows.length === 0) {
+      console.log('  No weekly cohort data found!');
+    } else {
+      console.log('\n  Week Starting    | Signups | Unique Users');
+      console.log('  ' + '-'.repeat(50));
+      weeklyCohorts.rows.forEach(row => {
+        const week = row.cohort_week.toISOString().split('T')[0];
+        console.log(`  ${week} |    ${row.signups.toString().padStart(4)} |         ${row.unique_users.toString().padStart(4)}`);
+      });
+    }
+
+    // 4. Sample retention calculation
+    console.log('\n\n🔬 4. SAMPLE COHORT RETENTION (First 4 weeks)');
+    console.log('-'.repeat(80));
+    
+    const cohortRetention = await pool.query(`
+      WITH signups AS (
+        SELECT 
+          user_id,
+          MIN(DATE(created_at)) as signup_date
+        FROM audit_logs
+        WHERE log_type = 'registration_completed'
+          AND user_id IS NOT NULL
+        GROUP BY user_id
+      ),
+      weekly_cohorts AS (
+        SELECT 
+          DATE_TRUNC('week', signup_date) as cohort_week,
+          user_id,
+          signup_date
+        FROM signups
+      ),
+      user_activity AS (
+        SELECT 
+          wc.cohort_week,
+          wc.user_id,
+          wc.signup_date,
+          DATE(al.created_at) as activity_date,
+          al.created_at - wc.signup_date as time_since_signup
+        FROM weekly_cohorts wc
+        LEFT JOIN audit_logs al ON al.user_id = wc.user_id
+          AND DATE(al.created_at) >= wc.signup_date
+      ),
+      retention_calc AS (
+        SELECT 
+          cohort_week,
+          COUNT(DISTINCT user_id) as total_users,
+          COUNT(DISTINCT CASE 
+            WHEN time_since_signup >= INTERVAL '1 day' 
+            AND time_since_signup < INTERVAL '2 days' 
+            THEN user_id 
+          END) as d1_users,
+          COUNT(DISTINCT CASE 
+            WHEN time_since_signup >= INTERVAL '7 days' 
+            AND time_since_signup < INTERVAL '8 days' 
+            THEN user_id 
+          END) as d7_users,
+          COUNT(DISTINCT CASE 
+            WHEN time_since_signup >= INTERVAL '14 days' 
+            AND time_since_signup < INTERVAL '15 days' 
+            THEN user_id 
+          END) as d14_users,
+          COUNT(DISTINCT CASE 
+            WHEN time_since_signup >= INTERVAL '30 days' 
+            AND time_since_signup < INTERVAL '31 days' 
+            THEN user_id 
+          END) as d30_users
+        FROM user_activity
+        GROUP BY cohort_week
+      )
+      SELECT 
+        cohort_week,
+        total_users,
+        d1_users,
+        CASE WHEN total_users > 0 
+          THEN ROUND(100.0 * d1_users / total_users, 1) 
+          ELSE 0 
+        END as d1_pct,
+        d7_users,
+        CASE WHEN total_users > 0 
+          THEN ROUND(100.0 * d7_users / total_users, 1) 
+          ELSE 0 
+        END as d7_pct,
+        d14_users,
+        CASE WHEN total_users > 0 
+          THEN ROUND(100.0 * d14_users / total_users, 1) 
+          ELSE 0 
+        END as d14_pct,
+        d30_users,
+        CASE WHEN total_users > 0 
+          THEN ROUND(100.0 * d30_users / total_users, 1) 
+          ELSE 0 
+        END as d30_pct
+      FROM retention_calc
+      WHERE total_users > 0
+      ORDER BY cohort_week DESC
+      LIMIT 4
+    `);
+    
+    if (cohortRetention.rows.length === 0) {
+      console.log('  ❌ Cannot calculate cohort retention - insufficient data');
+    } else {
+      console.log('\n  Cohort Week  | Users | D1    | D1%  | D7    | D7%  | D14   | D14% | D30   | D30%');
+      console.log('  ' + '-'.repeat(90));
+      cohortRetention.rows.forEach(row => {
+        const week = row.cohort_week.toISOString().split('T')[0];
+        const users = row.total_users.toString().padStart(5);
+        const d1 = row.d1_users.toString().padStart(5);
+        const d1pct = `${row.d1_pct}%`.padStart(5);
+        const d7 = row.d7_users.toString().padStart(5);
+        const d7pct = `${row.d7_pct}%`.padStart(5);
+        const d14 = row.d14_users.toString().padStart(5);
+        const d14pct = `${row.d14_pct}%`.padStart(5);
+        const d30 = row.d30_users.toString().padStart(5);
+        const d30pct = `${row.d30_pct}%`.padStart(5);
+        console.log(`  ${week} | ${users} | ${d1} | ${d1pct} | ${d7} | ${d7pct} | ${d14} | ${d14pct} | ${d30} | ${d30pct}`);
+      });
+      
+      console.log('\n  ✅ Cohort retention calculation is WORKING!');
+    }
+
+    // 5. Total stats
+    console.log('\n\n📋 5. OVERALL STATISTICS');
+    console.log('-'.repeat(80));
+    
+    const totalStats = await pool.query(`
+      SELECT 
+        COUNT(DISTINCT user_id) as total_registered_users,
+        MIN(created_at) as first_registration,
+        MAX(created_at) as last_registration,
+        EXTRACT(DAY FROM MAX(created_at) - MIN(created_at)) as days_of_data
+      FROM audit_logs
+      WHERE log_type = 'registration_completed'
+        AND user_id IS NOT NULL
+    `);
+    
+    if (totalStats.rows.length > 0) {
+      const stats = totalStats.rows[0];
+      console.log(`\n  Total Registered Users: ${stats.total_registered_users}`);
+      console.log(`  First Registration: ${stats.first_registration?.toISOString().split('T')[0]}`);
+      console.log(`  Last Registration: ${stats.last_registration?.toISOString().split('T')[0]}`);
+      console.log(`  Days of Data: ${Math.floor(stats.days_of_data)}`);
+      
+      console.log('\n✅ COHORT RETENTION ANALYSIS IS READY TO BUILD!');
+      console.log('\n📊 Next steps:');
+      console.log('  1. Create /api/analytics/cohort-retention endpoint');
+      console.log('  2. Add UI component to AnalyticsDashboard');
+      console.log('  3. Add CSV export for community manager');
+    }
+
+  } catch (error) {
+    console.error('\n❌ Error analyzing data:', error.message);
+    console.error(error);
+  } finally {
+    console.log('\n' + '='.repeat(80));
+    console.log('\n');
+    await pool.end();
+  }
+}
+
+// Run analysis
+analyzeCohortData().catch(console.error);
+

package/scripts/analyze-cohort-data.js

@@ -0,0 +1,295 @@
+/**
+ * Cohort Retention Data Analysis
+ * 
+ * This script analyzes the audit_logs table to determine if we have
+ * sufficient data to build cohort retention reports.
+ * 
+ * Usage:
+ *   node scripts/analyze-cohort-data.js
+ */
+
+require('dotenv').config();
+const { pool } = require('../services/db');
+
+async function analyzeCohortData() {
+  console.log('\n🔍 COHORT RETENTION DATA ANALYSIS\n');
+  console.log('=' .repeat(80));
+  
+  try {
+    // 1. Check for registration events
+    console.log('\n📊 1. REGISTRATION EVENTS');
+    console.log('-'.repeat(80));
+    
+    const registrationEvents = await pool.query(`
+      SELECT 
+        log_type,
+        COUNT(*) as total_events,
+        COUNT(DISTINCT user_id) as unique_users,
+        MIN(created_at) as first_event,
+        MAX(created_at) as last_event
+      FROM audit_logs
+      WHERE log_type IN ('registration_started', 'registration_completed', 'registration_submitted')
+      GROUP BY log_type
+      ORDER BY total_events DESC
+    `);
+    
+    if (registrationEvents.rows.length === 0) {
+      console.log('❌ No registration events found in database!');
+      console.log('   Cannot build cohort analysis without registration data.');
+      return;
+    }
+    
+    console.log('\nRegistration Events Summary:');
+    registrationEvents.rows.forEach(row => {
+      console.log(`  ${row.log_type}:`);
+      console.log(`    - Total Events: ${row.total_events}`);
+      console.log(`    - Unique Users: ${row.unique_users}`);
+      console.log(`    - Date Range: ${row.first_event?.toISOString().split('T')[0]} to ${row.last_event?.toISOString().split('T')[0]}`);
+    });
+
+    // 2. Check date range and volume
+    console.log('\n\n📅 2. SIGNUP DATE DISTRIBUTION');
+    console.log('-'.repeat(80));
+    
+    const dailySignups = await pool.query(`
+      SELECT 
+        DATE(created_at) as signup_date,
+        COUNT(*) as signups,
+        COUNT(DISTINCT user_id) as unique_users
+      FROM audit_logs
+      WHERE log_type = 'registration_completed'
+        AND user_id IS NOT NULL
+      GROUP BY DATE(created_at)
+      ORDER BY signup_date DESC
+      LIMIT 30
+    `);
+    
+    console.log('\nLast 30 days of signups:');
+    if (dailySignups.rows.length === 0) {
+      console.log('  No signup data found!');
+    } else {
+      dailySignups.rows.forEach(row => {
+        console.log(`  ${row.signup_date.toISOString().split('T')[0]}: ${row.signups} signups (${row.unique_users} unique users)`);
+      });
+    }
+
+    // 3. Check for user activity patterns
+    console.log('\n\n👥 3. USER ACTIVITY PATTERNS (Sample)');
+    console.log('-'.repeat(80));
+    
+    const userActivity = await pool.query(`
+      WITH user_signups AS (
+        SELECT 
+          user_id,
+          MIN(created_at) as signup_date
+        FROM audit_logs
+        WHERE log_type = 'registration_completed'
+          AND user_id IS NOT NULL
+        GROUP BY user_id
+        LIMIT 5
+      ),
+      user_events AS (
+        SELECT 
+          us.user_id,
+          us.signup_date,
+          DATE(al.created_at) as activity_date,
+          COUNT(*) as events_count,
+          COUNT(DISTINCT al.log_type) as unique_event_types,
+          EXTRACT(DAY FROM MIN(al.created_at) - us.signup_date) as days_since_signup
+        FROM user_signups us
+        LEFT JOIN audit_logs al ON al.user_id = us.user_id
+          AND al.created_at >= us.signup_date
+        GROUP BY us.user_id, us.signup_date, DATE(al.created_at)
+        ORDER BY us.user_id, activity_date
+      )
+      SELECT * FROM user_events
+      ORDER BY user_id, activity_date
+      LIMIT 50
+    `);
+    
+    console.log('\nSample user activity (first 5 users):');
+    if (userActivity.rows.length === 0) {
+      console.log('  No user activity data found!');
+    } else {
+      let currentUser = null;
+      userActivity.rows.forEach(row => {
+        if (row.user_id !== currentUser) {
+          currentUser = row.user_id;
+          console.log(`\n  User: ${row.user_id.substring(0, 8)}...`);
+          console.log(`  Signup: ${row.signup_date.toISOString().split('T')[0]}`);
+        }
+        console.log(`    Day ${row.days_since_signup}: ${row.events_count} events (${row.unique_event_types} types)`);
+      });
+    }
+
+    // 4. Check for referral/source tracking
+    console.log('\n\n🎯 4. SOURCE/REFERRAL TRACKING');
+    console.log('-'.repeat(80));
+    
+    const sourceTracking = await pool.query(`
+      SELECT 
+        metadata->>'hasReferralCode' as has_referral,
+        metadata->>'referralCode' as referral_code,
+        COUNT(*) as count
+      FROM audit_logs
+      WHERE log_type = 'registration_completed'
+        AND user_id IS NOT NULL
+      GROUP BY metadata->>'hasReferralCode', metadata->>'referralCode'
+      ORDER BY count DESC
+      LIMIT 10
+    `);
+    
+    console.log('\nReferral tracking in registrations:');
+    if (sourceTracking.rows.length === 0) {
+      console.log('  No referral data found in metadata');
+    } else {
+      sourceTracking.rows.forEach(row => {
+        console.log(`  Has Referral: ${row.has_referral || 'unknown'}, Code: ${row.referral_code || 'none'} - ${row.count} users`);
+      });
+    }
+
+    // 5. Test cohort query structure
+    console.log('\n\n🔬 5. COHORT RETENTION FEASIBILITY TEST');
+    console.log('-'.repeat(80));
+    
+    const cohortTest = await pool.query(`
+      WITH signups AS (
+        -- Get first registration event per user
+        SELECT 
+          user_id,
+          MIN(DATE(created_at)) as signup_date
+        FROM audit_logs
+        WHERE log_type = 'registration_completed'
+          AND user_id IS NOT NULL
+        GROUP BY user_id
+      ),
+      weekly_cohorts AS (
+        -- Group signups into weekly cohorts
+        SELECT 
+          DATE_TRUNC('week', signup_date) as cohort_week,
+          user_id,
+          signup_date
+        FROM signups
+      ),
+      user_activity AS (
+        -- Get all user activity after signup
+        SELECT 
+          wc.cohort_week,
+          wc.user_id,
+          wc.signup_date,
+          DATE(al.created_at) as activity_date,
+          al.created_at - wc.signup_date as time_since_signup
+        FROM weekly_cohorts wc
+        LEFT JOIN audit_logs al ON al.user_id = wc.user_id
+          AND DATE(al.created_at) >= wc.signup_date
+      ),
+      retention_calc AS (
+        -- Calculate retention metrics
+        SELECT 
+          cohort_week,
+          COUNT(DISTINCT user_id) as total_users,
+          COUNT(DISTINCT CASE 
+            WHEN time_since_signup >= INTERVAL '1 day' 
+            AND time_since_signup < INTERVAL '2 days' 
+            THEN user_id 
+          END) as d1_users,
+          COUNT(DISTINCT CASE 
+            WHEN time_since_signup >= INTERVAL '7 days' 
+            AND time_since_signup < INTERVAL '8 days' 
+            THEN user_id 
+          END) as d7_users
+        FROM user_activity
+        GROUP BY cohort_week
+      )
+      SELECT 
+        cohort_week,
+        total_users,
+        d1_users,
+        CASE WHEN total_users > 0 
+          THEN ROUND(100.0 * d1_users / total_users, 1) 
+          ELSE 0 
+        END as d1_pct,
+        d7_users,
+        CASE WHEN total_users > 0 
+          THEN ROUND(100.0 * d7_users / total_users, 1) 
+          ELSE 0 
+        END as d7_pct
+      FROM retention_calc
+      WHERE total_users > 0
+      ORDER BY cohort_week DESC
+      LIMIT 10
+    `);
+    
+    console.log('\nSample cohort retention (last 10 weeks):');
+    if (cohortTest.rows.length === 0) {
+      console.log('  ❌ Cannot calculate cohort retention - insufficient data');
+    } else {
+      console.log('\n  Cohort Week      | Signups | D1 Users | D1% | D7 Users | D7%');
+      console.log('  ' + '-'.repeat(70));
+      cohortTest.rows.forEach(row => {
+        const week = row.cohort_week.toISOString().split('T')[0];
+        const signups = row.total_users.toString().padStart(7);
+        const d1 = row.d1_users.toString().padStart(8);
+        const d1pct = `${row.d1_pct}%`.padStart(4);
+        const d7 = row.d7_users.toString().padStart(8);
+        const d7pct = `${row.d7_pct}%`.padStart(4);
+        console.log(`  ${week} | ${signups} | ${d1} | ${d1pct} | ${d7} | ${d7pct}`);
+      });
+      
+      console.log('\n  ✅ Cohort retention calculation is FEASIBLE!');
+    }
+
+    // 6. Summary and recommendations
+    console.log('\n\n📋 6. SUMMARY & RECOMMENDATIONS');
+    console.log('-'.repeat(80));
+    
+    const totalSignups = registrationEvents.rows.find(r => r.log_type === 'registration_completed');
+    const oldestSignup = dailySignups.rows[dailySignups.rows.length - 1];
+    const newestSignup = dailySignups.rows[0];
+    
+    if (totalSignups && oldestSignup && newestSignup) {
+      console.log('\n✅ DATA AVAILABILITY:');
+      console.log(`  - Total Signups: ${totalSignups.total_events} events, ${totalSignups.unique_users} users`);
+      console.log(`  - Date Range: ${oldestSignup.signup_date.toISOString().split('T')[0]} to ${newestSignup.signup_date.toISOString().split('T')[0]}`);
+      console.log(`  - Cohort Analysis: POSSIBLE`);
+      
+      console.log('\n📊 RECOMMENDED COHORT GROUPINGS:');
+      console.log('  - Weekly cohorts (best for initial analysis)');
+      console.log('  - Monthly cohorts (for long-term trends)');
+      console.log('  - Daily cohorts (if high volume)');
+      
+      console.log('\n📈 METRICS WE CAN TRACK:');
+      console.log('  ✅ signup_cohort (week/month starting)');
+      console.log('  ✅ signup_date_range (date range string)');
+      console.log('  ✅ total_signups (count)');
+      console.log('  ✅ d1_users, d1_pct (day 1 retention)');
+      console.log('  ✅ d7_users, d7_pct (day 7 retention)');
+      console.log('  ✅ d14_users, d14_pct (day 14 retention)');
+      console.log('  ✅ d30_users, d30_pct (day 30 retention)');
+      console.log(sourceTracking.rows.length > 0 ? '  ✅ source (from referral tracking)' : '  ⚠️  source (limited/no tracking)');
+      
+      console.log('\n🚀 NEXT STEPS:');
+      console.log('  1. Create /api/analytics/cohort-retention endpoint');
+      console.log('  2. Support weekly and monthly cohort grouping');
+      console.log('  3. Add source/referral filter if available');
+      console.log('  4. Build visualization in AnalyticsDashboard');
+      console.log('  5. Export to CSV for community manager');
+    } else {
+      console.log('\n❌ INSUFFICIENT DATA:');
+      console.log('  Cannot build cohort retention without registration events');
+      console.log('  Ensure registration_completed events are being tracked');
+    }
+
+  } catch (error) {
+    console.error('\n❌ Error analyzing data:', error.message);
+    console.error(error);
+  } finally {
+    console.log('\n' + '='.repeat(80));
+    console.log('\n');
+    await pool.end();
+  }
+}
+
+// Run analysis
+analyzeCohortData().catch(console.error);
+

package/scripts/analyze-prod-cohorts.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+# Analyze cohort retention data on production Heroku database
+
+echo "🔍 Analyzing cohort retention data on dubs-server-prod..."
+echo ""
+
+# Get the DATABASE_URL from Heroku and run the analysis
+heroku config:get DATABASE_URL --app dubs-server-prod | \
+  xargs -I {} env DATABASE_URL={} node scripts/analyze-cohort-data-heroku.js
+