VK has migrated its data analytics operations to a proprietary unified technology platform. The system integrates data storage, search, quality control, access management, reporting, and product experimentation tools. Currently, 90% of the company's analysts utilize this infrastructure, which manages a data volume exceeding 400 petabytes.
The platform centralizes data from VKontakte, VK Video, Max, and Odnoklassniki. It supports various workflows, including ML model development and reporting. According to company representatives, the automation of these processes has doubled the speed of decision-making and reduced computational resource consumption by 60,000 CPU cores.
A dedicated module for product experimentation tracks methodologies and results, enabling teams to reuse data and test hypotheses more efficiently. The system also standardizes reporting metrics across all analytical functions, which minimizes manual labor and ensures consistent data-driven decision-making throughout the organization.
The platform facilitates approximately 1,000 A/B tests on a monthly basis. By providing a unified approach to data handling, the company aims to accelerate product development and improve the accuracy of user scenario analysis.
