Startups

Exacaster
2014
Apr 03

Exacaster offers advanced software platforms, tools and machine-learning algorithms for customer churn prediction, up-sell targeting, product recommendations, segmentation and real time dynamic pricing.

We use open-source Hadoop stack with Spark as the core platform for data processing, and develop our proprietary algorithms in this environment. Our focus is on complete automation of the customer relationship management area in Telecommunications and customer Loyalty programs in Retail using machine learning as the core decision maker to replace middle-level managers.

We have developed automated customer retention, upsell and promotion processes where the entire process all the way from data ingestion, then segmentation, then predictive modeling, then campaign execution and finally ROI evaluation is machine-learning driven and automated, forming a complete, closed observe-decide-act learning loop. Today we touch data on nearly 40 million consumers every day across Europe, US and Latin America with deployments among leading companies in Telco and Retail.

We work with Oozie, Hive, Impala, Hbase, Kafka, Python, Spark and some proprietary ML frameworks. We are currently researching applications of deep learning algorithms to our problem domain and will commercialize results in 2017.

We leverage nearly complete sets of all internal enterprise data consisting of transactons/sales receipts, call records, mobile data records, network level data, CRM activity data, web browsing data and more, plus external data which can be obtained in the market such as mobile device profiles or location demographical profiles. These data sets are the merged into comprehensive customer 360 degree views which serve the as the basis for further analysis and automation.