Managing reputational risk is a prime concern in the banking industry. Banks go to great lengths to secure their reputation. However, as history suggests, risks can come suddenly and seemingly out of nowhere taking everyone by surprise. This explains the rise of regulatory oversight in the form of Basel III, the Sarbanes-Oxley Act, Fair Debt Collection Practices Act, Home Mortgage Disclosure Act, Truth in Lending Act, etc. Regulations help - but only to an extent. But banks, the healthcare industry and just about any business that touches consumers can see damage being inflicted from social media as well. In such cases, there is little that regulation can do to insulate business from irreparable damage.
In China, Sina Weibo - a Twitter-like service, recently showed us the kind of havoc ordinary consumers can create. In an incident where a customer of a rural bank made a request to withdraw 200,000 yuan, the request was turned down. The incident was reported on Sina Weibo, leading to a run on the bank from panic stricken account holders who thought the bank was actually going bankrupt while it wasn't.
In such a situation, what could the bank have done to address the negative information on the social media channel? The bank would have done well to put in place systems that identify and proactively address such incidents in context - even taking into consideration - unsubstantiated and unproven information that is likely to affect business. Reactive crisis management is history in the banking business and every other industry as well.
The trick is to listen to what customers are saying - continuously. Then, sift through the data using advanced analytics to "discovery" insights that could have a negative impact on business. The concept, called data discovery, is central to modern reputation management. The key goal of such a data discovery engine would be to accelerate time-to-insight using analytics, pre-built industry-specific applications and dashboards with advanced visualization that enable accurate and quick decision making.
Data discovery engines work towards making sense of the vast volume of data from sources that include social media, blogs, comments, email, chats, text messages, surveys, branch check ins, ERP, POS, ATMs, CRM, etc. Much of this real time data is unstructured and the engine must be architected to:
- Automatically detect, surface and contextualize negative content in the least possible time
- Prioritize comments / reviews based on severity and sentiment
- Respond to placate / assure customers before the problem/ issue goes viral
The science behind a data discovery engine is different from how traditional EDW are organized and managed. Here, there is no pre-defined 'schema' for the data. Everything is just pushed into a "data lake". The data is extracted and loaded based on the industry-specific need and analytical models/ algorithms being used. It is then enriched by coupling incidents with suggested corrective action. End result: a highly effective methodology to proactively identify and respond to risk.
The difference between traditional methodologies and a data discovery engine needs to be emphasized. The first largely depends on knowing the kind of risks that exist and then looking for them within data. Sometimes this works; sometimes - as the bank in China discovered - there is no time to analyze the data. It is already too late. The second methodology ensures that you are never "too late".