CLV Forecast model

What is predicted CLV?

Predicted CLV (Customer Lifetime Value) is the estimated total revenue LiftSignals expects from a customer over the next 12 months. This is different from historical LTV (what they've spent so far) — it's a forward-looking prediction based on their behaviour patterns.

How the model works

LiftSignals uses a probabilistic model (based on the Pareto/NBD and Gamma-Gamma frameworks) combined with gradient-boosted features from your store. Inputs include:

• Historical purchase frequency and recency
• Average order value and trend direction
• Product category affinity
• Seasonal patterns from same-cohort customers
• Current churn risk score (declining engagement reduces CLV forecast)

CLV tiers

LiftSignals automatically segments customers into four CLV tiers based on their predicted 12-month value relative to your store's distribution:

Why is my CLV forecast lower than expected?

The CLV forecast is personalised to each customer and reflects their actual purchase trajectory. Common reasons for lower-than-expected forecasts:

• The customer's purchase frequency has been declining (increasing churn risk)
• Missing first_order_date data limits the model's ability to understand their tenure
• The customer is in the At-Risk or Critical churn band — declining engagement reduces the forecast
• Fewer than 2 orders (the model has limited data to work with)