Machine learning and artificial intelligence have transitioned from the promise of the future to an established tool in the credit decisioning environment in the last few years.
One financial services executive told me recently that 70% of their models used some form of ML or AI. Using an algorithm to manage customers was inconceivable some years ago, whereas now, it’s the only way to process complex volumes of data.
Think of the data you have about customers in your own company. If you have current account or credit card transaction data, you will be able to profile a customer’s spending and derive insights from that data.
The rollout of Open Banking has made even greater volumes of data available, so transaction data isn’t just restricted to a customer’s bank, providing they consent to share the information. Instead, it’s data that can be plugged into decisioning engines and used to understand customers better.
In the coming years, it’s increasingly likely that more new data sources will become accepted in the financial services industry. Social media data may be used to help understand customers who have little or no traditional financial track record, while we are already seeing web data insights used to aid decisions about small businesses.
A key hurdle for implementing new data sources and the Ml and AI technology needed to process and harness it is ensuring the systems can be explained to regulators and customers alike. How your decisioning engine uses data and arrives at decisions on customers needs to be fair and transparent.
It’s a challenge we have anticipated. Experian’s PowerCurve Strategy Management provides the capability to deploy AI and ML models today in a way that the regulator and customer needs for explainability can be met.
The next task is to progress new models from development through the pilot stage and, finally, implementation. As more organisations become accustomed to managing data, this process is likely to speed up. There could be a competitive advantage for those who identify highly predictive new data sources early.
Once the structure is in place, the size of the data lake will only grow, with more unstructured data being used as part of decision-making processes.
It’s important to remember that merely holding data is of little use – it’s the analytics to unlock its insights that have true value.
For more information, watch Richard Topham in conversation with Armando Capone here