Productivity Gains through Chat and Email Analysis
To illustrate the variety of application of Context Intelligence we will publish in the coming weeks various case studies.
Our second case demonstrates how with Context Intelligence banks are able to catch escalation scenarios substantially earlier generating substantial savings while at the same time increasing client satisfaction.
Company: A global financial firm with large foreign exchange trading (Forex) activities. The bank is reputed to have a best-of-class Forex-trading platform and wants to extend this industry leadership to the monitoring and analysis of the forex trades. Because of regulatory demands, the fast paced nature of the Forex business and legacy systems, an actual trade may be passed between up to 12 different systems (systems for trading, ledger, front-, middle-, back office, etc.).
Challenge: The bank lacks at any one point in time a 360-degree view on the condition of its Forex operations. Last year it started to implement a solution combining the structured data into one single stream for analysis in a BI platform. However exceptions and escalations often are best traced in the chat and emails accompanying a trade. These types of information are typical cases of unstructured information and have so far not been incorporated in any analysis. However, trading exceptions and escalations – e.g. when a trade did not close at the projected price because of market movements – are sources of disputes and of considerable costs. The bank sought to close the loop by incorporating the analysis of chat and email data to get a true 360-degree view of their Forex operations.
Figure 1 – Interface for Chat Analysis
Solution: To capture the structured elements of any Forex transaction (trade, log entries, transaction times, etc.) the bank deploys a state of the art bank specific ETL-framework to deliver the necessary data input to a BI solution for analysis. Squirro provides the missing piece – analysis of unstructured data. Loading and analysing chat and email, automatically analysing the data on a number of dimensions such as exception, escalation, and compliance issues, the results are pushed to the BI layer for in depth exploration. The in-built alerting function of Squirro allows triggering events programmatically when certain escalation conditions are met.
Context generation: The analysis of the chat and email data streams is based on state-of-the-art techniques such as density based spatial clustering, topic evolution analysis, and cross-document reference and concept detection. The resulting patterns are normalized and pushed to the BI platform.
Results: By applying context intelligence techniques the bank is able to generate a full 360-degree view of the health of their Forex operations. They are able to catch escalation scenarios substantially earlier saving the bank massively while at the same time increasing client satisfaction. Return on investment calculations show cost savings of > $1 million with ROI figures of 25-30% per annum.