Rabobank Learned Valuable Big Data Lessons Thanks to Proof of Concepts
Rabobank started with big data in March 2011. After they named the big data trend one of 10 most important trends in their IBA ICT Trends prediction. In the prediction of 2013 trends, the big data trend was positioned at number 6 of ICT trends.
As part of practice what you preach, the Rabobank started with developing a big data strategy in July 2011. They created a list of 67 possible big data use case , which they wanted to investigate further. These use cases included:
- To signal and predict risks the bank is running;
- To signal, detect and prevent fraudulent actions;
- To identify customer behaviour;
- To obtain a 360-degrees customer profile;
- To recognize the most influential customers as well as their network;
- To be able to analyse mortgages;
- To identify the channel of choice for each customer.
These uses cases, and all others, were divided into four main categories:
- To fix organisational bottlenecks;
- To improve efficiency in business processes;
- To create new business opportunities;
- To develop new business models.
For each of these categories they roughly calculated the ICT impact, time needed to implement it as well as the value proposition. In the end the Rabobank moved forward with big data application to improve business processes as they required less ICT impact and had the possibility for a positive ROI.
Rabobank started with a few proof of concepts (POCs) and they first started using only internal data. Next to internal data, Rabobank distinguishes internet data (click behaviour), social data (from social networks), public data (from government sources) and trend data.
In order to be able to test several big data tools for different use cases, Rabobank decided to build a small Hadoop cluster. This clusters consisted of 16 nodes including 1 master node. Three 128 Terabyte servers using Hive, Pig and MapReduce were used to analyse high-level unstructured data sets. The Hadoop cluster was installed at Rabobank’s data centre in The Netherlands, however it was placed in a dedicated production environment.
A dedicated, highly skilled and a multidisciplinary team was created to start with the big data use cases. The culture among the team members was important for the success of the POCs. In order to stay up to speed, they worked with small and short cycles and most importantly it was allowed to make mistakes as long as the mistakes provided a learning experience. After the small use cases, the objective is to move on to more complex cases.
Use Cases Rabobank
One of the use cases was to create an auto-complete function for mobile banking. With this feature, users would not have to use their address book anymore. Instead, the system would auto-complete account information when a user types an account number. Of course, it should not be possible to view account information of unknown people. Therefore, the system analysed 3 billion transactions in the financial network. When a search history of 14 months was used, 99% of the accounts had 122 or less unique contra accounts. Using only 6 months of transaction history, this would be 79 contra accounts or less. Thanks to this big data tool, mobile banking has become a lot more customer friendly.
Another use case of the Rabobank was to analyse criminal activities at ATMs. Rabobank found out that the proximity of highways, the season and weather condition increased the risk of criminal activities. Harrie Vollaard, innovation manager at Rabobank, explains that they also used big data to analyse customer data to find the best places for ATMs.
According to Harrie Vollaard, creating a big data strategy is not easy and eventually this should be an important part of the overall strategy of the bank. Therefore, Rabobank is working on further developing the big data competence centre, combining people with various backgrounds that can broaden big data at Rabobank.
The small POCs allowed Rabobank to learn valuable lessons from big data. According to Marcel Kuil and Hilde Hulten, they found out that the big data technology is ready and not expensive to implement when open-source tools are used. The Hadoop cluster that they used delivers high performance with low costs and can be scaled linearly. For Rabobank, the key to success was the multidisciplinary team and that they embraced uncertainties and accepted mistakes to be made.
There were also challenges. The privacy issues will be a concern in the future and Hadoop is only part of a big data strategy. At the moment the Rabobank does not store raw data, due to the costs and capacity issues. The data quality was not constant and the security issues have to be very high. During the process the Rabobank noticed that it was often unclear who owned the data as well as were all data was stored. Finally, they noticed that specialised knowledge as well as visualizations are very important to drive big data success.
Since the board of Rabobank is now familiar with big data, there are more important big data subjects on the agenda. Items like data ownership, privacy issues as well as what consumers still consider service are now being discussed. With big data everything is possible, but the questions is whether that makes customers automatically happy.