Churn prediction
How Skubl identifies customers at risk of churning before they leave.
Skubl's churn prediction model scores every customer in your database with a risk probability. Customers are flagged as high-risk when their behaviour deviates from their personal purchase cadence.
Signals the model uses
- Days since last purchase vs. personal average purchase interval
- Decline in order frequency or average order value
- Engagement drop across email or on-site behaviour (if Klaviyo or GA4 is connected)
- Cohort comparison — how this customer compares to similar customers at the same tenure
Acting on churn risk
Review the Churn Risk page
Go to Customers → Churn Risk to see the list sorted by risk score.
Create a segment
Click 'Create segment from high-risk' to build an audience of at-risk customers.
Activate the segment
Push the segment to Klaviyo for a win-back email sequence, or to Meta Ads for a retargeting campaign.
A churn prediction is most useful when it's early. Set up an alert in Intelligence → Alerts to be notified whenever a segment's churn risk score crosses a threshold.
Related articles
Was this helpful?
If something's unclear or missing, get in touch and we'll update the docs.