Parker Fitzgerald (PFG) were selected by UBS bank as their trusted delivery partner for a pioneering project to prove the value and practical application of data driven predictive analytics.
The objective was to harness data, modelling and cost-effective technical design and implementation to deliver an innovative, forward-looking approach to the identification and management of operational risks. Whilst similar techniques had been applied in more traditional spheres of risk management (credit, market), it was ground-breaking and potentially transformative to apply the thinking to the identification and management of risks around operational resilience.
Our approach using a data-driven machine learning model, hosted on a cost-effective technology platform, sought to disrupt the traditional manual and retrospective approach to managing such risks. A priority use case around global technology change risk was selected to demonstrate the value, scalability and extensibility of the approach, whilst maximising the immediate return on investment for the client.
The key challenges faced were the sheer scale of global technology change across the organisation, the complexity of identifying and proactively mitigating the underlying causes of IT change related incidents, and the finite capacity of existing technology management and support teams to manage the risks.
We delivered an optimised model that leveraged a diverse set of internal data sources to identify key risk themes and to predict the probability of failure for future IT change events, hosted on a technical solution based on open source components for ease of support and future extension.
The key benefits delivered to the client were:
• Automated insights into correlated risks and ‘toxic combinations’ around IT Change;
• A forward-looking view of emerging and correlated risks;
• A prediction of the probability of failure for specific IT changes, weighted towards the most severe business impacts; and
• A solution that could be extended and scaled non-linearly, for sustainable support and cost of ownership.
Testing against historic data evidenced that the model was able to provide early warning of >30% of change related failure incidents, providing a clear financial business case for the project investment. Embedding the data driven and forward-looking model outputs into UBS’ technology risk management practices, was designed to foster a culture of ‘prevention rather than cure’ and enable effective targeting of finite technology budget and resources.