Customer Churn Calculation

Our analytics platform uses a data-driven approach to calculate customer churn based on historical order patterns. This ensures churn definitions align with your actual customer behavior rather than arbitrary timeframes.

How Churn is Calculated​

The system analyzes the distribution of days between orders for your customers to establish intelligent thresholds for "at risk" and "churned" statuses.

Order Distribution Analysis​

The system automatically:

1. Collects historical data on days between consecutive customer orders
2. Creates a statistical distribution of these intervals
3. Identifies key percentile points in this distribution
4. Uses these percentiles to set default churn thresholds

Key Metrics​

The churn calculation uses two primary thresholds:

At Risk Days​

The "At Risk" threshold identifies customers who haven't placed an order in a time period that suggests they may be at risk of churning. By default, this is set at the 40th percentile of your order distribution.

For example, if your data shows:

• 40% of your repeat orders happen within 29 days
• 60% of repeat orders take longer than 29 days

Then customers who haven't ordered in 29+ days would be flagged as "at risk."

Churn Days​

The "Churned" threshold identifies customers who haven't placed an order in a time period that suggests they have likely abandoned your business. By default, this is set at the 80th percentile of your order distribution.

For example, if your data shows:

• 80% of your repeat orders happen within 47 days
• Only 20% of repeat orders take longer than 47 days

Then customers who haven't ordered in 47+ days would be classified as "churned."

Example Distribution​

Here's an example of what your order distribution might look like:

Customizing Churn Settings​

While the system provides data-driven defaults, you can customize these thresholds to match your business needs:

Adjusting At Risk Threshold​

You can set a custom number of days after which a customer is considered "at risk" of churning. This may be useful if:

• Your sales cycle has seasonal variations
• You have subscription-based products with specific billing cycles
• You want to be more proactive in customer retention

Adjusting Churn Threshold​

You can set a custom number of days after which a customer is considered "churned." Consider adjusting this if:

• Your business has long sales cycles
• Your products have specific replacement timeframes
• Your customer base includes both frequent and infrequent shoppers

Best Practices​

1. Review your distribution data before making changes to understand your natural order patterns
2. Start with the data-driven defaults before making adjustments
3. Consider your business model when customizing thresholds
4. Test different thresholds to find the optimal balance between:
   • Identifying truly at-risk customers
   • Not generating too many false positives
   • Catching churn early enough for effective intervention
5. Regularly review and update your churn settings as your business evolves

Technical Implementation​

The system:

1. Retrieves your store's days-between-orders curve via API
2. Calculates the statistical distribution of days between orders
3. Identifies key percentile values
4. Sets default thresholds at P40 (at risk) and P80 (churned)
5. Respects any custom thresholds you've already configured

Any changes to these thresholds are saved in your store settings and applied to all churn calculations going forward.