Case Study: Active Super improves churn prediction with Deepend's AI solution

Active Super boosts churn prediction accuracy to 75% with Deepend

Active Super, a prominent Australian superannuation fund, faced the significant challenge of customer churn, a costly issue in an industry with low barriers to switching. They partnered with their long-term technology vendor, Deepend, to leverage artificial intelligence in a new way. The challenge was to predict which members were likely to leave by analyzing vast amounts of siloed and unstructured data, including customer service interactions.

Deepend engineered a custom data pipeline and developed an AI application using the LangChain framework and OpenAI's APIs. This solution was designed to identify churn signals in both demographic and sentiment data and to generate personalized retention communications. The initial results were promising, with churn prediction accuracy exceeding initial targets by 5% and subsequent fine-tuning by Deepend increasing accuracy to 75% in testing rounds.

Active Super