Takahiro Noda, Data Analyst& Data Management and Yoshihisa Kabumoto, Senior Data Architect, IT Strategy & Transformation, AXA
Big data is an important enabler to achieve business objectives.To activate big data opportunities, it is recommended to set up an environment which is not limited to information technology but also encompasses business strategy, human side, and culture. In this article, we introduce key concepts that lead you to success with big data.
Data Strategy: What you should do first for utilizing big data is neither installing a Hadoop cluster nor hiring smart data scientists. You need to start with finding where big data fits in your business. The findings should be materialized as a data strategy, which should identify its key challenges with prioritized actions in alignment with own business target. This data strategy enables stakeholders to capture new opportunities offered by big data, and ensure a comprehensive collection of decisions, leaderships, and data sources that support target businesses.
Data Management: Outcomes of big data analysis rely on input data quality. The “garbagein, garbage out” rule is stillalive in big data environments. In order to ensure that the data quality is standardized and maintained, it is mandatory to establish a data governance body and proceed to implement data management practices. Under the supervision of the data governance body, data stewards should conduct data standardization, validation, and reconciliation processes. Besides, enterprise metadata management aligned with data governance is another key point. By defining and utilizing metadata, we can monitor lifecycles of data, and our data scientists can easily and quickly identify appropriate data set from a vast amount of data sources.
Data Organization: Big data is also a company asset and should not be monopolized by a single line of business. Data science skills, culture, and capabilities should also be shared within the enterprise. To create an effectivemulti-disciplinaryteam, and to avoid building a new silo, you should consider to form a cross-organizational data team, which should be organized by members from various organizations, such as IT, Distribution, Marketing, Product Development and Finance. Activating local capabilities and the talent pool on the market, you can choose the best mix of analysis team members at any big data projects.
Infrastructure: From technology viewpoint, big data is open or inexpensive in general. Software for big data is often open-sourced and hardware is highly commoditized. However there is an exception to this. Keeping big data infrastructure stable and available is cumbersome and expensive. If you can take advantage of your global team and organization, it is advisable to set up a shared platform. By using this shared private infrastructure, you can maximize cost efficiencies for developing and operating big data infrastructure, and protect security, privacy, and confidentiality of customers’ data, since those technologies are typically all sharable across companies.
Social Exemplarity: Security and privacy are major concern for customers.To ease such concerns, you should clearly state your transparent data usage in the interest of customers and society, and depict compliance standards in line with both regulatory constraints and your own data ethics.
Big data tend to be thought as a matter of technology. However, it covers a wide range of subjects around data such as data strategy, data management, a data organization, and infrastructure. The last piece to success with big data is continuous transformation as both market and technology around big data evolves moving forward.