Powering Custella’s Marketing Analytics: How Alteryx Works Behind the Scenes

Expansion in cashless market increases burden in reporting tasks

Sumitomo Mitsui Card Company is working to “realize a better cashless society” through solutions for members, affiliated stores and corporations, beginning with issuing credit cards and providing payment terminals. The company offers “Custella,” a service that leverages cashless data on one of the largest scales in Japan, including credit card data, to help companies understand customer profiles and provide marketing support.

Mr. Toru Ishikawa, who oversees data analysts at the company’s Data Development Headquarters, explains, “Custella is a comprehensive marketing support service that statistically utilizes attribute information such as gender, age, place of residence, annual income and family structure obtained during credit card applications, along with payment data including who made payments where, and for how much. The service performs everything from corporate customer profiling to efficacy verification. It has been widely adopted by both local governments and companies, providing detailed analysis reports that include consumer behavior, and offering consistent services that extend to direct promotions targeting specific customers.”

The current cashless market is rapidly expanding, with transaction data growing from 8 million transactions per month in 1999 to over 500 million transactions per month by 2024, a more than 60x increase. Notably, there was an approximately 2.5x increase in the five years from 2019 to 2024. Consequently, the company’s data processing load has become significantly heavier.

The rapid growth of cashless data has resulted in the creation of more than 200 reports per year for Custella’s services. The shortest report took two to three weeks to produce, and the longest took six months or more.

Further, challenges including an increase in types of data handled (including external data), increased members and expanded data volume that can’t be managed conventionally in Excel or PowerPoint became apparent. To address these challenges, the company started implementing Alteryx with three users in 2022 and is now utilizing it across the entire organization.

Standardizing analytics while empowering every analyst

With the introduction of Alteryx, the company achieved significant results in two primary areas.

The first area was “improving the efficiency of data inspection.” During the course of many analysis projects, when there was a sense of inconsistency with analysis results, further investigation revealed reports of “errors in data extraction.” Since humans are involved in the work, individual differences in the quality of scrutiny arose, and human error, such as overlooked check items, became a challenge.

“By leveraging Alteryx, we have defined inspection workflows and can now automatically generate graphed PDF reports by simply inputting data. This allows anyone to analyze data from a consistent perspective, and since the entire process from data input to output is completed in 2 to 3 minutes, we have minimized potential human error to the extent possible, irrespective of operator level of experience,” explains Mr. Ishikawa.

The second area was the “democratization of machine learning.” There were different situations when machine learning, such as decision tree analysis and clustering analysis, was necessary, but challenges arose due to the need for preparatory measures, establishing execution environments and individual skill discrepancy. Particularly in financial institutions, the barrier to installing software on PCs from an information security perspective was high, and although Python was the desired programming language at the time, it was difficult to implement.

Mr. Ishikawa adds, “With Alteryx, by streamlining everything from pre-processing to actual execution, anyone can now handle machine learning and achieve high reproducibility. Decision tree analysis can also be completed in about 10 minutes from initial flow. This makes it easier to first apply decision tree analysis, allowing for easy pattern extraction.”

Mr. Ishikawa also states that when leveraging Alteryx, it is essential to compare it with other tools to appropriately deploy it depending on the situation.

He offers further detail: “For small- to medium-sized projects, we use ETL tools from other companies, while Alteryx is employed for professional features and handling large-scale data. Specifically, Alteryx excels in its wide range of data input/output formats, connectivity with different databases, support for diverse file formats such as SAS, Hyper and Shape files, and the capability to output to PDF without switching tools. Additionally, Alteryx offers unique advantages like the ability to check intermediate outputs at any time and control certain processes by enclosing them in a square frame using the container function, providing analysts with flexible control.”

Sumitomo Mitsui Card Company promotes the use of Alteryx for the entire organization, preparing learning materials and conducting basic training in line with its enterprise. It’s encouraging further use of Alteryx by adopting an approach that gradually progresses from aggregation with sample data to application with in-house data. Further, improvements in learning efficiency have also been achieved by combining Alteryx with generative AI.

Mr. Ishikawa details, “I believe Alteryx’s learning is highly compatible with generative AI. I also sensed that the Japanese documentation is still lacking in some areas, and generative AI has significantly supplemented these gaps.”

Regarding future developments, the company aims to achieve further value creation through ongoing technological innovation and collaboration with AI.

Mr. Ishikawa says, “To continuously generate value from data, we believe it is crucial not only to further expand data but also to enhance the sophistication of technology including tools, machine power and algorithms that efficiently derive insights, and to integrate these with human resources. We aim to improve the quality of human decision-making by promoting collaboration between humans and AI, utilizing Alteryx in the process.”