Transforming the way people manage their taxes

Navigating the complexities of data silos

It’s no secret that the combination of data and AI technology is revolutionizing the tax market. Not only does data technology make it possible to systematically collect and analyze users’ various tax information such as personal income and expenses, it has also dramatically simplified the execution of complex tax and accounting tasks such as estimating expense types more efficiently. As a leading innovator in the Korean tax market, Jobis & Villains successfully introduced a new paradigm in the domestic tax market, utilizing data and AI to simplify how taxes are done through automated expense processing classification, personalized communications with customers, and data-driven tax strategy services.

As the digitization of tax and accounting services increased, so did the volume of data and the complexity to process and analyze it. Reliant on their own databases to perform business analytics, Jobis & Villains met with many data management and utilization challenges. The siloed data sources meant data needed to be processed in different pipelines, making it difficult to manage data in a unified way. The lack of efficient data loading also hindered data governance and collaboration within the data team. Even when using the same source data, metrics were extracted differently by different analysts, while the same metrics were created with different names, making it difficult to explore the data efficiently.

Additionally, data that could not be processed in the existing AWS Glue catalog could only be viewed in other environments, and downstream analysis performance was unstable due to inconsistent data extraction criteria. With data coming from multiple channels and through multiple means of extraction, it was difficult to leverage business intelligence (BI) to make data-driven decisions. Furthermore, due to the nature of the tax business, large amounts of data are generated in a given period, making it difficult to collect and manage efficiently. Due to these data challenges, Jobis & Villains chose the Databricks Data Intelligence Platform to unify all their data in a centralized manner — simplifying data management, exploration, analytics and visualization. Taeho Cho, a data engineer at Jobis & Villains, mentioned, "Data democratization was critical for us to achieve our goals for driving a data-driven culture, and Databricks was the most suitable solution." 

Democratizing unified data with Databricks Data Intelligence Platform

Turning to Databricks as the foundation for their data and analytics initiatives, Jobis & Villains has been able to unify their previously scattered data and feed them to unified pipelines based on data characteristics for downstream analytics. With Unity Catalog, they have also implemented unified data governance and controls across the company. They are now able to manage data delivery channels in a standardized way and leverage Databricks SQL to efficiently extract and analyze metrics from their common data. This has also made data validation easier in many ways.

In addition to the original data, secondary data can also be utilized in Unity Catalog, significantly increasing data reuse. “Databricks has enabled us to analyze the original data from multiple angles,” said Jinsoo Park, a data analyst at Jobis & Villains. Since Jobis & Villains actively utilizes Databricks as a collaboration platform, it’s easier than ever for their teams to share data and insights with others in the company — by loading company-wide data into the lakehouse and leveraging powerful features within the interactive notebooks, such as data descriptions and commenting, and management of data pipelines in the public catalog.

These newly established data workflows have increased the business’ efficiency since employees from each department can analyze aspects of their business using common metrics. For example, prior to Databricks, the SamjumSam platform was using inconsistent metrics related to its mobile service, such as monthly active users, sales amount, payment amount and the number of payments with different aggregations for each department. After implementing the Databricks Platform, teams are able to share data across the company, enabling all team members to obtain a unified view of the data to determine business direction, resulting in improved alignment.

Expanding data initiatives to support more diverse business opportunities

Since Databricks has enabled Jobis & Villains to quickly aggregate and share large amounts of data with their team, they have successfully democratized data and increased business efficiency in the process. Now, with data-driven decision-making at the helm of all business processes, the company is empowered by their newfound ability to explore more diverse opportunities to grow the business. For example, before Databricks, it was not possible to make real-time decisions about customer behavior and take specific actions. After building a real-time streaming pipeline on the Databricks Platform, they are now able to analyze streaming data and provide insights to improve interactions instantly. This has led to the development of a data workflow that sends real-time prompts to users when they leave a certain area within their platform. As a result of these targeted interactions, the company has generated 2 billion KRW in additional sales. “Databricks has helped us to do things that we couldn’t do before,” Cho emphasized.

Plus, due to the nature of taxes, the SamjumSam platform sees the most client visits in the month of May, with over 300 million user behavior events. Since Jobis & Villains can utilize the right resources at the right time, Databricks standardized data sources and pipelines make it easy to analyze large volumes of data cost-effectively when tax season arrives. “With Databricks, we can process massive volumes of data at a very fast rate,” said Cho. Park added to Cho’s comment, ”Since we can consolidate our vast data set in a centralized manner, we can use it to analyze various indicators and set goals. This will allow us to expand our business beyond the tax market.”

Most importantly, Databricks has significantly improved data engineering efficiency and resource utilization. In other words, managing multiple pipelines once consumed significant engineering resources to build and maintain, but with Databricks, all pipelines can be managed with a unified solution. Monitoring and other automation tasks are also easily handled by the fully managed platform — reducing data engineer resources to manage pipelines by 20%. The team is now able to manage more data pipelines with fewer people, and, in turn, more resources can be allocated to new data platform tasks and future initiatives. “With Databricks, we can utilize our data engineers’ resources more diversely,” concluded Cho. Due to the success of using the Databricks Platform, Jobis & Villains will continue to lead the tax market’s charge toward digitization — using both data and AI as their guiding light. 

Moving into the future, Jobis & Villians plans to continue leading their digitization efforts with data and AI, and expand beyond the tax market — with Databricks as a key enabler.