Standard Chartered Bank - Building a Collaborative Solution for Data Transformation and Portfolio Recommendations
Name: Sathya Sivam Sivaraj, Head, Business Analytics and Intelligence
Organization: Standard Chartered Bank
Standard Chartered Bank in Singapore is part of a leading international banking group, with a presence in 59 of the world’s most dynamic markets for more than 160 years and serving clients in a further 83. Our purpose is to drive commerce and prosperity through our unique diversity, and our heritage and values are expressed in our brand promise, here for good. For more information, please visit www.sc.com/sg.
Value at Scale
Most Impactful Transformation Story
Relationship Managers (RM) engaging Wealth Management Clients need up-to-date portfolio details of the client for a productive engagement, to present an accurate picture of which investment avenues are making a profit, and to impart relevant investment advice.
RMs use a lot of upfront manual effort to collate relevant data from disparate systems, perform portfolio returns analysis, and identify top-performing assets. This is a time-consuming process and limits the number of client engagements.
Another acute challenge is engaging in ad hoc discussions with clients. Market volatility is prompting more and more clients to seek frequent ad hoc advice from respective RMs, which necessitates readily accessible portfolio details and analytics. Manual portfolio analysis is non-standard and subject to the personal preferences of RMs. Multiple attempts to standardize portfolio data collation and relevant analytics yielded short-term standardization benefits. Because of the subjective elements in portfolio analytics, change or turnover of RMs caused disruption in continuous and steady client servicing and affected overall client satisfaction.
While all relevant data is available and can be manually collected, the key issue was collating data in a consistent manner, homogenizing data from disparate systems, standardizing portfolio analytics, and providing a minimal manual effort. Off-the-shelf and consultative solutions are available in the market but were prohibitively overpriced; in one market an automated solution that partly addresses the need for a consolidated portfolio was quoted as high as a million dollars with commensurate AMC. We needed a way to provide accessible portfolio analytics in a cost-effective manner.
We used Dataiku not only to automate portfolio analytics with data from disparate systems but also to extend to the next logical step of recommending portfolio rebalancing to maximize ROI, tailored to the different risk appetites of the clients!
Collating data across sources is very easy in Dataiku. With all relevant data available in one place, next, we homogenize data elements to normalize similar data sets. This led to the next logical set of validating the quality of source data.
Leveraging Dataiku’s built-in features, the quality of incoming data was assessed and recorded. Business rules for data curation were added in Dataiku to address data gaps and other quality issues. Standardized and curated data sets were then transformed as expected by Portfolio Managers.
Creating an intuitive, informative, and accurate portfolio in a BI tool was made simple as we had ready-made aggregated data with drill-down to individual transactional details. Clients’ portfolios are available to RM at the click of a button. Producing client portfolios was very responsive as the BI tool just fetched relevant data readily made available by Dataiku.
With Dataiku's built-in models, client investments across different asset classes are assessed against the Bank’s Model Portfolio to help clients capture global opportunities while managing risk exposure. The platform also considers investments that are adding the least value to the client in portfolio reallocation recommendations.
Along with the client’s portfolio, RMs have investment recommendations, details of the top and worst performing assets, and trend analytics. These were not possible in a manual process.
RMs are able to have wholesome investment conversations with clients and add value by providing quicker and richer insights. With rebalancing recommendations personalized to clients’ risk appetites, RMs are now focusing on client engagement rather than upfront preparation.
Value Brought by Dataiku:
When assessing the quality of source data with automated data set metrics and computation in Dataiku, the DQ is expressed as a single score. Recording this made the addition of DQ analytics to end users possible; the DQ dashboard improved business trust in transformed data. At the same time, it provides timely alerts from Dataiku based on thresholds on DQ score signaled issues upfront, increasing transparency on data reliability, and ensuring timely and systematic fixes in source systems.
With the ease of handling partition data in Dataiku, accumulating portfolio details across periods is made very easy. Performing trend analysis on the accumulated data added further value to portfolio analysis and for modeling personalized investment recommendations.
Transformation logic, data quality assessment, and data curation need proper and thorough documentation. Typically, data transformation and documentation reside in different locations. Dataiku’s built-in Wiki brings the two together. With the ability to add hyperlinks to data sets and recipes, it is easy to document business and other transformational logic, organize them meaningfully, and present them structurally.
All the above are definitely not available in any other solution available in the market. This solution was built with a team of three focusing on sourcing and data transformation, portfolio recommendations, and presentation respectively!
Dataiku makes collaborative work simple, and it was easy to work on the different parts of the solution with ease. With a turnaround time of three months, the overall cost was considerably less but added a lot more value to RMs.