New Model Improves Revenue Forecasting by Analyzing Customer Behavior

Kyeongbin Kim, Daniel McCarthy, Dokyun Lee· August 5, 2026 View original

Key takeaways

  • Disaggregating revenue into customer-base drivers significantly improves forecasting accuracy.
  • The CBMT model reduces forecasting error by 30% over traditional benchmarks.
  • Coordinated forecasting is most effective when customer acquisition, retention, and spending patterns are interdependent.
  • Granular forecasts provide actionable insights for optimizing marketing and sales strategies.

Who benefits

RetailE-commerceSaaSFinancial ServicesTelecommunications

Summary

Researchers developed the Customer-Based Multi-task Transformer (CBMT) to forecast revenue by disaggregating it into customer acquisition, repeat purchasing, and spending per order, significantly outperforming existing benchmarks. The model learns shared structures across these drivers while retaining separate forecasts, aligning them for overall revenue prediction.

A new research paper introduces the Customer-Based Multi-task Transformer (CBMT), an advanced model designed to enhance revenue forecasting accuracy. Unlike traditional aggregate forecasts, CBMT breaks down revenue into its fundamental customer-base drivers: new customer acquisition, repeat purchases, and average spending per order. By simultaneously learning shared patterns across these drivers and maintaining individual forecasts, the model provides a more granular and insightful prediction. Evaluated across 966 companies in 25 industries, CBMT demonstrated a 30% reduction in mean total-sales error compared to leading customer-base benchmarks. It also showed a slight, though not statistically significant, improvement over direct total sales forecasting Transformers. The model's ability to coordinate these customer-base forecasts proves particularly beneficial for firms where these primitives exhibit strong co-movement, offering a clearer picture for revenue planning.

Why it matters

This research offers a more accurate and granular approach to revenue forecasting, enabling businesses to better understand the underlying drivers of their sales and make more informed decisions on budgets and demand planning.

How to implement this in your domain

  1. 1Evaluate current revenue forecasting models for their ability to disaggregate customer-base drivers.
  2. 2Explore integrating multi-task learning architectures to simultaneously predict acquisition, retention, and spending.
  3. 3Analyze the co-movement of customer-base primitives within your business to identify potential gains from coordinated forecasting.
  4. 4Pilot the CBMT methodology on a subset of products or customer segments to assess its accuracy and insights.
  5. 5Adjust marketing and sales strategies based on the granular forecasts to optimize customer acquisition and retention efforts.

Original post by Kyeongbin Kim, Daniel McCarthy, Dokyun Lee

"arXiv:2608.02911v1 Announce Type: new Abstract: Revenue forecasts guide acquisition budgets, demand planning, and customer-based valuations, yet an aggregate forecast does not show whether change reflects acquisition, repeat purchasing, spending per order, or offsetting movements…"

View on X

Originally posted by Kyeongbin Kim, Daniel McCarthy, Dokyun Lee on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses