Chief Technology Officer, Open Lending.
In lending, good data can be the difference between profit and failure. Knowing your borrower and the landscape leads to smarter decisions about who to lend to and how much to lend at which rate.
Data is arguably the most valuable resource across industries. Getting existing data in a healthy place to inform AI tools and other decisions is top of mind for many business leaders moving into 2024, but it is only the first step toward data governance.
Implementing the right policies is critical for capturing good, clean data and encourages people to stop number-crunching and start taking advantage of the AI and ML tools at their fingertips. However, to use these tools effectively, it’s important to first ensure quality data and then understand how to effectively process it.
As you prepare for 2024, keep these three data Qs in mind to build a data strategy that can carry your business through the era of AI.
Quality Of Data
Data quality is non-negotiable if you plan to use an AI or ML tool in your business. However, these tools will only generate helpful, trustworthy results if your foundational data is accurate and complete. Without a solid foundation to stand on, they might as well be making up solutions. That does not help anyone.
Your own data is hardly the limit for AI and ML tools. You should also consider industry data, economic data, customer data and maybe even competitor data. In lending, you should further consider actuarial and federated data to maintain ownership and control. All these outside sources must also meet data quality standards to contribute to your results meaningfully.
The world has come a long way from large Excel file data dumps. Spreadsheets have evolved into data warehouses that standardize the data and make it consumable for analytics and reporting, but there is still lots of room for misinterpretation. A data governance committee is crucial to create and enforce policies that encourage quality data collection and cleanliness.
Quantity Of Data
Data collection should be top of mind for every business planning to use AI or ML now or in the future. Many businesses are planning to do just that. Ensure you have a clear data collection policy outlined in your company charter and customer agreements so that you are not limited in what you can achieve by lack of access to data.
Collecting as much data as possible will provide more variables for an AI or ML tool to analyze. The more data provided, the more accurate the results will be—not because every data point will factor into every decision but because having a more robust dataset will allow you to drill down and discover the individual data points that matter.
For example, I once worked with a lender dealing with a substantially different loss curve than was common. At first, we could not pinpoint the cause, but after 18 months of data collection, we eventually had enough loan data to find the point of commonality among the failing loans. Once identified, we could adjust their program to correct the issue and improve overall losses. Imagine what the result would have been if we had not collected that one singular data point. We would still be scratching our heads to this day!
Quality Of Partner
The odds are that you are not managing this influx of data alone. You do not need to be an AI or ML expert to take advantage of these incredibly effective tools as partnerships can be made to ensure success. This might take the form of an external partner or internal expertise.
In loan decisioning, several sets of key data come together to provide the decision and terms of a potential loan. Borrower data, industry data and risk factors are all contributing factors. A quality partner will use these three datasets to make connections that predict the probability and severity of default instead of relying exclusively on a borrower’s creditworthiness—a common mistake in loan decisioning.
Relying on traditional measures of creditworthiness alone could lead a lender to inadvertently offer a loan with a high payment-to-income ratio. Providing a high-risk loan like this sets the lender and the borrower up for failure. A good partner will save you from that failure by making connections within existing datasets to better determine if a loan will be profitable. By analyzing alternative attributes such as the number of payments made or credit inquiries over the past year alongside potential loss, lenders can choose a term or loan structure that is a better fit for their portfolio and their customers.
Creating A Comprehensive Data Strategy
A holistic data strategy will connect a quantity of quality data with a quality data partner to provide important business insights in the age of AI and ML. As these tools become more pervasive and useful, businesses without a solid data foundation may risk falling behind. Now is the time to reflect on these three data Qs to understand how they might strengthen your position ahead of the AI evolution.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
Source link
