DEV Community

Cover image for A Case Study of Tableau’s Revolutionary Influence in the BFSI Domain
Taashee Linux Services
Taashee Linux Services

Posted on • Originally published at taashee.com

A Case Study of Tableau’s Revolutionary Influence in the BFSI Domain

HDFC Bank is the largest private sector bank in India by assets and the 10th largest bank in the world by market capitalization, with 97% of their branches, customers, and assets located in the country. They are present across the length and breadth of India and currently have 68 million customers. Every year, they acquire about 2-3 million customers, and as of now, they have a total of 6300+ branches. One of the unique things about the bank is that about 50% of its branches are located in the deeper geographies of India, which poses a different set of challenges when it comes to connectivity. Overall, HDFC Bank is a large and prominent player in the Indian banking industry.
It is worth noting that HDFC Bank is a relatively young bank, having been established in 1995. Their businesses are broken up into three segments: retail, wholesale, and treasury. Within retail banking, they offer a wide range of loan products, including microfinance loans as low as $150, as well as funding for local small businesses, large government corporates, and the government itself. They are also the second largest collector of taxes for the government of India. On the wholesale side, they cover everything from project finance to merchant banking and advisory services.
Both the wholesale and retail segments currently use Tableau.
HDFC and Tableau through the years
The Reserve Bank of India (RBI) is the regulator for HDFC Bank, and they have a conservative approach to freedom. To mirror this, the bank has created a Business Intelligence (BI) unit that mirrors the structure of the business itself. This includes a BI team for wholesale, one for retail, and one for the treasury. Within retail, there are different segments for assets and liabilities. This centralizes all performance management data and makes the bank responsible for managing and analysing it.
This approach is different from others where data management and analysis are handled by separate teams. Having a single unit for performance management ensures that data governance and access to data are controlled and secure, which is particularly important for a bank with 1,50,000+ employees and a high risk of data breaches.
How it all started
In 2015, HDFC Bank was looking to replace Excel with a more efficient tool for managing and analysing performance data. They were introduced to Tableau through Taashee, a Tableau Power Analytics Partner, and after a successful proof of concept, they decided to implement the software. However, they didn’t take Tableau to the masses straightaway, instead, they started with a smaller group of employees, specifically the wholesale unit. This was a small data set, but it was deep and complex, allowing them to test the capabilities of Tableau.
They used Tableau to create a monthly account profitability report, which gave them a three-month pattern of data. This allowed them to switch off Excel and gain insights that they wouldn’t have been able to see before. A number of employees were still more comfortable with Excel reports, but HDFC was able to use Tableau to create databases from the reports and generate further reports from them, reducing employee workload by a large margin. As they started implementing Tableau, they also rolled out their first Tableau desktop license, which allowed employees to access the software from their own devices.
So, the need to remove Excel served as the primary driver for the implementation. With Tableau, HDFC took a significant step in terms of managing and analysing performance data. They were able to visualize data much easier, get a better understanding of the performance of different business units, and identify any issues that needed to be addressed.
After implementing Tableau in the wholesale unit, HDFC Bank decided to expand its use to the SME segment as well. In 2016, as a result, they acquired an additional 80 crore in SME customers, which increased the size of their customer base by ten times. The intention was to also roll out Tableau to the retail segment, but this was not realized till 2017.
Challenges and breakthroughs during Tableau implementation
After implementing Tableau, HDFC Bank was able to significantly reduce its turnaround time for performing data analysis. This was due to the use of sequential databases which allowed them to quickly crunch data and produce effective, easily decipherable reports.
Before 2015, the process of generating reports was time-consuming and relied on multiple people to perform different tasks, such as roll-ups and data cuts. The reports were generated in Excel sheets and then sent via email, which led to delays. With Tableau, the process became much faster and more secure. They were able to produce a monthly account profitability snapshot, which gave them an overview of the performance of the wholesale unit.
The implementation of Tableau also allowed them to handle ad-hoc requests from wholesale customers more efficiently. Before Tableau, customers would frequently ask for data points such as maturity profiles or LC information, which would add to the workload of the business intelligence team. With Tableau, they were able to anticipate these requests and create a digital wholesale databank, which reduced their ad-hoc workload and allowed them to focus on more important tasks.
For those who are implementing Tableau for the first time, one of the key to-do’s is to get the single sign-on feature right. This was something that HDFC Bank implemented in 2016 with the help of the Tableau team. It reduced friction and made it easier for employees to access the software. In the previous version, employees had to log into windows, email, and Tableau separately. With single sign-on, they were able to access all three with one click.
With the increase in Tableau adoption throughout the difference, they found that they were able to reduce the number of reports they sent out. They now only send out three reports: the API, which is a monthly report, the ALM, which is a business position report that gets refreshed daily, and a majority report, which helps people stay informed about the performance of different business units.
The daily ALM report rolls up data from the contract level. The report is fed by the core banking system on a daily basis and includes data on all different products such as loans, bills, and overdrafts. The report includes data on the customer level along with company-level data, which helps identify which customers are picking up which products.
They added a radio button feature that allows users to drill down and see the outstanding balance and movements that have occurred over a period of time. There are also a bunch of filters at the top of the dashboard that allows the supervisory layer to be more effective.
The bank also added a maturity profile feature, which gives customers an overview of what will happen if no action is taken. It provides data on different assets, liabilities, and off-balance sheet items. The report is masked to show only superficial values, but the numbers behind it are actual data collected over time. This feature allows for an informative snapshot without the risk of someone breathing down on the bank for sharing sensitive information.
The bank has developed a system that uses colour coding to organize company details, which helps their teams plan their outreach meetings and tasks more effectively. Additionally, they have customized the system for specific business sets, such as for a new customer who requested a new design. By having a set of data ready, they are able to quickly spin a new layout, and go back and ask their users if the design is working for them, as the design is ultimately done by the Tableau team.
As they progressed through the next phase of implementation, they were able to get 10,000 users on the Tableau server. To ensure that the speed of the analysis remain uncompromised, they started extracting data from a p7 Oracle box and placing it in a centralized, regulated environment. This allowed for more efficient analysis, as it was not done manually by the ground-level employees.
In terms of organization, the business team first determines the customer requirements and then comes to the BI team for analysis. They also have an internal finance team that figures out what needs to be done with the customer data. As they strive to improve their data analysis, they have been focusing on governance to ensure that data is not floating around and that there are not too many people analysing it.
Bringing Tableau’s innovations to the branch level
HDFC bank faced a number of challenges while implementing its data analysis system for the retail sector. In 2017, the bank started working on this project, but faced issues such as poor connectivity in rural branches and multiple different hierarchies to manage, compared to their wholesale sector. Taashee’s Tableau engineers were actively involved in this phase of the implementation.
One of the main challenges that the bank faced was the lack of bandwidth for pushing data to the branches’ performance sheets. The product team used to take data from the business intelligence unit and present it to the branches, but the formats and timing of the data were inconsistent. To solve this issue, the bank took eight months to build the data and establish a consistent journey for the branches. Another challenge was the issue of deep geography, where the branches in remote areas faced buffering issues when trying to access the data. The bank solved this by enabling subscriptions and sending the data to the branches through an email channel at 8.30 am in the morning before the branches began operations.
Taashee worked with the bank to further create a branch view on Tableau, that is accessible to the branches through an email channel, which includes various dashboards for different products such as auto loans, credit cards, and others. The data is generally refreshed on a monthly basis, but some parameters are refreshed daily. This high-level view of the branches’ performance allows the branches to figure out the parameters they need to focus on since the bank has set periodical targets for the number of new accounts to be opened and the amount to be maintained in these new accounts for each branch. These targets are tracked and monitored through a deeper dashboard that provides a 12-month trend, month-to-date position, and resource-level information. The branch supervisors can view the percentage of branches meeting the plan, while the branch managers can view the percentage of resources meeting the plan.
One of the unique features of the bank’s retail sector is the profitable niche of managed customers, which includes 10 million+ customers, of which 3.5 million+ are managed by physical relationship managers (RMs) and the rest by virtual RMs. The bank has created scorecards and insights for these customers on Tableau, which give a summary of the different programs running, how business heads are contributing, and a month-on-month summary of value creation or erosion.
The branches further requested a solution that would allow them to act quickly and efficiently to rectify their performance levels. To address this, the bank has developed a Tableau dashboard which is specifically designed for use in India, where a customer can have two different types of accounts –savings and current. The dashboard allows branch managers to see what happened the previous day, whether money flowed out of the branches or not. This is particularly important in today’s age of internet banking, where money can be transferred outside of the bank at any time, and the branch manager may not be aware of it until it is too late. The bank uses multiple hierarchies to group the data points differently, and the dashboard has the ability to drill down to the customer level. This is admittedly a large set of data, but thanks to Tableau, the bank was processing 44 million records overnight (as of 2017) and making them available by 8:30 am each morning before the branches opened.
All these features that have developed over time as per requirements have now been collated into an integrated Tableau dashboard. It is a simple tool that shows on the customer level, which customers have a checking account, a current account, a savings account, a salaried relationship or a fixed deposit or time deposit with the bank and the status of such accounts. At the bank level, it displays information about managed and unmanaged customers, as well as their location (metro, urban, semi-urban, rural) to understand how different customers and products interact. One of the key features of the integrated dashboard is the ability to change targets and perform what-if analyses, which has been critical in changing the way the bank approaches data. It provides information about the top and bottom performers at the branch level, allows for cross-hierarchy comparison, and has roll radio buttons allowing users to quickly tap through and change the dashboard, making it easy for users to understand and compare different products.
The path forward
In recent years, branch managers, particularly the new ones, have provided feedback that there is an overwhelming amount of data available in the sheets being generated, making it difficult to know which sheets to access for specific information. The goal for both HDFC and Tableau now is to create a simple, natural language AI interface with a chat function, that allows managers to easily access the information they need.
The bank has scaled out its technology over time, starting with an 8-core system in 2015 and moving to 16 cores. They added more cores over time without changing the architecture itself. The bank currently is at 96 cores and has recently signed up for another 84, bringing the total to 176.
As per the latest reports, HDFC had 26,500+ users on the Tableau server, 200+ desktop licenses and had also deployed Tableau on Linux!
This article was originally published on our company blog.

Top comments (0)