Global direct-to-consumer slashes membership churn

A direct-to-consumer food and beverage service was experiencing unexpected customer churn for its subscription-based beverage program.

freight boxes loaded in aircraft



overall reduction 

in churn


would-be churn

detected in advance

1, 3, 6

month campaigns made from predictions

The challenge

Identify and treat unexpected customer churn

As a provider of subscription-based beverage plans that ships internationally, this company saw an unexpected rise in subscription churn but was unable to detect why—and to predict who might churn in the future. The company was engaging in ineffective and generic re-engagement campaigns which failed to stymie the churn, caused a negative customer experience from a lack of targeting, and yielded poor campaign ROI.

In addition to determining why customers churn and who was likely to churn in the future, the company wanted to effectively target high-risk customers as far as 3 months in advance.


Automatically prep and unify data from 48 beverage lines

With the Pecan platform, this company automatically connected to and unified data across 48 beverage lines. This data was comprised of marketing campaigns, distribution data, CRM activities, promotions, and more.

In just days, the client had their fully rained retention model that prioritized VIP subscribers. This model enabled the marketing team to provide a preemptive retention strategy by segment, time, and customer value, and feed these predictions directly to their CRM.


85% of churn detected and treated in 1, 3 and 6 month campaigns

The Pecan platform was able to identify 85% of customer churn in advance which enabled the marketing team to create focused campaigns 1, 3, and 6 months in advance resulting in vastly improved customer retention and lifetime value.

The company saw an 11% overall reduction in subscription churn, and reduced their surge in churn they were experiencing by 39%. For VIP customers, the overall churn rate was lowered by 20%.

Want to get started with predictive analytics?