Reactivating lapsed and lost customers for a premium coffee chain

Our client, one of the world’s largest coffeehouse chains, was losing revenue due to a decline in recurring customers.

Before TransOrg’s intervention, active customers or customers purchasing in the last 90 days were contributing 17% of the total revenues. However, client wanted to increase the share of revenues by active customers to 25% of its total revenues.

TransOrg implemented its advanced analytics and machine learning based solutions and analyzed lapsed customers (transacting between 90 to 180 days ago) and lost customers (transacting more than 180 days ago) to reactivate high potential and high probable customers by targeting them with relevant promotion offers.

TransOrg consolidated, normalized, and harmonized sales data from multiple data sources and created a clean master DataMart that was further used for analytics.

For the analysis, TransOrg segregated the non-transacting customers into two distinct groups of lapsed and lost customers based on their most recent transaction dates. Further, with feature engineering TransOrg created additional variables such as total spend, average transaction size (AT), number of transactions, and promotion flags to further segment customers in each group based on promotion redemption and transaction spend behavior.

TransOrg created seven sub-segments within both the lapsed and lost customer groups, namely:

No promotion redeemed:

1. Single transactions
2. Single transactions (with only birthday benefits only)
3. More than a single transaction with AT<600
4. More than a single transaction with AT>600

Promotion redeemed:

1. Single transaction
2. 2 to 3 transactions
3. more than 3 transactions

TransOrg further analyzed which promotion offers should be offered to each sub-segment.

Key Impacts

• Increased the recurring customer base from 17% to 25% of the total customer base.
• Converted lapsed and lost customers to active customers through personalized
  promotion offers while increasing total sales.