Enhanced analytics – or better decision-making information – are one of the key goals of data transformation. However, it is a goal that financial services firms continue to struggle to achieve. Panellists at the recent A-Team Group Data Management Summit in New York City discussed challenges the industry faces and potential approaches to getting analytics correct and complete.
The panel, ‘Unlocking value – laying the foundations for smart analytics’, was joined by Peter Kapur, managing partner, True North Data Consulting; Shailesh Mathankar, director, data management, AIG; Edgar Zalite, global head of metadata management, chief data and innovation office, Deutsche Bank; and Bhushan Bhangale, director, head of enterprise data quality remediation and analytics, BNP Paribas.
Facing significant analytics challenges
Financial services firms are struggling to put in place better data management to support smart analytics for a number of reasons. First, the industry has historically been an early adopter of technology, which means that there are a lot of outdated systems in many firms that are hard to integrate, or harvest data for analytics from.
As well, much of this historic software adoption was driven by a specific business process outcome, and not built from the data up. As a result, data sits in silos determined by business process or business line, or even a specific application. It can be hard to compare apples-to-apples.
Another challenge that firms are facing is geographic and regulatory. Large financial firms with global footprints have to contend with data regulations in each specific country in which they operate, and this can often make data management challenging. For example, in Switzerland, data cannot be moved outside the country’s borders. In Russia, encrypted data isn’t permitted within the country. Rules like these make it difficult to integrate data across the silos they create to develop analytics.
The growth of new data sets also raises the bar. Employees within firms want to use new data sets, including alternative data. However, existing data management structures can struggle to accommodate these new types of data and transform them into analytics.
All this can mean that data management teams spend a lot of time trying to figure out how to drive analytics from existing practices and processes, and engage a lot of ML – not machine learning, but manual labour.
The obvious solution, to adopt a data governance framework supported by a technology platform, can prove challenging in financial services business lines that have short-term targets to meet. Often, data teams are not viewed as business-friendly because initial data governance projects focused on compliance issues such as BCBS 239 or anti-money laundering model validation. Data teams can be viewed as policemen, not partners.
Creating the change that’s required
Cultural change is needed to alter these terms of engagement. Panellists suggested that data teams:
- Look at the data already in data governance systems and consider how it might be leveraged to create analytics the business would find useful.
- Speak with the business and find out where its important business analytics challenges are. Think about how data management might solve these.
- Work through how data management resources are supplied to the business. If data management supplies a data lake, what are the requirements the business must adhere to to use it? How can the data lake better support the business in creating analytics – for example, with a good data catalog?
- Find people to sit within the data management team who really understand the business, and vice versa, people in the business who understand data. Individuals who are bi-lingual in this way have a better chance of uncovering the right business analytics challenges that can be solved through better data approaches.
- Be flexible – there may be a part of the business that wants or needs to use a different data governance platform, or analytics toolkit, than the central one adopted. Understand the reasons why this is the case and if it makes sense, support this alternative approach.
- Showcase the analytics capabilities that the CDO office and the data governance programme have to the business on a regular basis. Educate the business about past successes, and what could be achieved.
- Seek out the parts of the business that are innovating with analytics and engage with them. Demonstrate the value that good quality data can bring to artificial intelligence (AI) and machine learning (ML) projects. If necessary, bring on additional data science expertise.
- Help the business uncover the right external data sources to work with – for example, third parties that consistently provide good quality data.
- Support self-service. At the end of the day, in most organisations the business still owns the data. So, any approaches should enable the business to work with analytics on its own and not insert data governance as a gatekeeper.
- When beginning a new compliance-driven project, think about how the data could be repurposed later to support business analytics.
In short, there are many ways in which the data management team and the office of the CDO, can support the development of smart analytics in partnership with the business. Getting the business and the data teams to communicate and collaborate better is often a good step in the right direction.