Fitness Studio Member Retention Engine

How It Works

1. 1
   Data Extraction

   Connects to Mindbody API using OAuth credentials and pulls attendance records, class bookings, and member profiles for the previous 7-day period, filtering for active members only

   Mindbody REST API
2. 2
   Pattern Analysis

   Calculates key metrics including attendance frequency decline, booking-to-show ratios, class type diversity, and time-since-last-visit for each member using rolling 4-week averages

   Python pandas
3. 3
   Risk Assessment

   Sends structured member data to Claude AI with prompts designed to identify churn risk patterns, requesting numerical scores and specific behavioral flags for each member

   Claude API
4. 4
   Risk Flagging

   Processes Claude's analysis to identify members with churn scores above configurable thresholds (default 70%), categorizing them into risk levels and storing results in tracking database

   SQLite database
5. 5
   Alert Trigger

   Automatically triggers Win-Back Campaigner for newly flagged at-risk members and generates summary reports for studio management via email notifications

   SMTP email

Implementation

# Churn Risk Scanner Implementation

## File Structure
```
churn_scanner/
├── main.py
├── mindbody_client.py
├── claude_analyzer.py
├── database.py
├── config/
│   └── settings.json
├── templates/
│   └── alert_email.html
└── logs/
```

## Core Functions

### main.py
```python
import logging
from datetime import datetime, timedelta
from mindbody_client import MindbodyClient
from claude_analyzer import ChurnAnalyzer
from database import Database

def run_churn_analysis():
    # Initialize connections
    mb_client = MindbodyClient()
    analyzer = ChurnAnalyzer()
    db = Database()
    
    # Pull weekly data
    end_date = datetime.now()
    start_date = end_date - timedelta(days=7)
    attendance_data = mb_client.get_attendance(start_date, end_date)
    member_profiles = mb_client.get_member_profiles()
    
    # Analyze patterns
    risk_scores = analyzer.analyze_churn_risk(attendance_data, member_profiles)
    
    # Flag at-risk members
    flagged_members = [m for m in risk_scores if m['risk_score'] > 0.7]
    
    # Store results and trigger campaigns
    db.store_risk_scores(risk_scores)
    trigger_winback_campaigns(flagged_members)
```

## Environment Variables
```
MINDBODY_API_KEY=your_mindbody_api_key
MINDBODY_SITE_ID=your_site_id
CLAUDE_API_KEY=your_claude_api_key
DATABASE_URL=sqlite:///churn_analysis.db
SMTP_SERVER=smtp.gmail.com
SMTP_PORT=587
ALERT_EMAIL=management@peakstudios.com
```

## Cron Setup
```bash
# Add to crontab for Monday 7am execution
0 7 * * 1 /usr/bin/python3 /opt/churn_scanner/main.py >> /var/log/churn_scanner.log 2>&1
```

Data Flow

Prerequisites

• -Mindbody API credentials with read access to attendance and member data
• -Claude API key with sufficient monthly token allocation
• -Python 3.8+ environment with pandas, requests, and sqlite3 libraries
• -Cron service configured and running on host system
• -SMTP email credentials for alert notifications

Error Handling