A gaming industry client was looking to better understand the balance of performance for their in-game products across their player base. A combination of large, hard-to-manage volumes of data, as well as unstructured proprietary datasets, made extracting signals [to best predict anomalies in product performance] a monumental task for their internal resources.
We developed a bespoke data ontology tool and custom ETL processes in order to understand the latent connections across data in the organization, and with this tool constructed a real-time anomaly detection model that highlighted not only product imbalances, but the signals in each product that could have contributed to the performance change.
The data ontology tool provided the client’s data teams with a method to efficiently query their big data as well as extract more insightful signals on the engagement and play style of their customers. The anomaly detection model provided the product managers, business managers, and executives of the organization with a novel method for identifying the balance of performance for products, and concurrently democratized an aspect of the industry that was historically limited to subject matter experts.