Buy now, pay later (BNPL) is a major trend in e-commerce. Seen as an alternative to traditional credit cards, the concept of BNPL is not new, but technology has revolutionized the process of evaluating and extending instant consumer credit for both online retailers and third-party applications. And it's catching on.
Financial services industry research and consulting firm Cornerstone Advisors estimates that consumers made more than $24 billion in online BNPL purchases in 2020, and that in 2021 the number will be about $100 billion. That's a significant portion of the $933 billion e-commerce market, and it stands to grow even bigger as the organizations involved in extending instant credit to consumers become better at measuring risk. And that is where things get tricky.
Consumer credit is a moving target, and it can be hard for a BNPL service to make a precise measurement of a particular person's credit risk profile. Information that might influence a credit rating one way or the other could be difficult for the service to access, data may be incomplete, or it could be obsolete by the time it is used in a risk calculation. That's bad for both the consumer and the lender.
Extending credit to consumers who may not be able to pay back their loan is not a good business practice for a lender. What is good business is to extend enough credit that the customer benefits from the convenience, justifying whatever fees or interest payments are associated with the service. The question is, how can a lender arrive at the best decision fast enough to keep pace with the speed of the market, and with the kind of accuracy that maximizes the potential of each customer?
That's hard to do because financial data is sensitive, and protected by a number of state and federal regulations. Information like mortgage and credit card account balances and interest, auto loans, and even seasonal spending patterns could be used to measure risk and support better decision making. But accessing and using that data to run the models necessary to generate predictive intelligence is risky.
Traditionally, putting that data to work meant transferring it from where it is stored to the environment where the predictive models would be run, then decrypting it to extract the desired predictions. Every move and every moment that personally identifiable information is not encrypted increases risk, and may even violate security and privacy laws.
The risk equation complicates matters for those organizations that want to take advantage of cloud service providers. They get caught between the economic benefits of cloud-based storage and compute, and the mandate to keep data protected. Once data is encrypted and moved to the cloud, it is often stuck there, unused.
Using Cape Privacy's powerful combination of secure multiparty computation and secret sharing, financial services organizations can operationalize their most sensitive consumer data and run their preferred AI prediction models to gain a more precise understanding of their customers–without ever decrypting their data. And through its partnership with data cloud operator Snowflake, Cape Privacy streamlines the processes involved by allowing firms to collect and encrypt their data, move it into the Snowflake data cloud, and run those AI prediction models in Snowflake while remaining encrypted.
The alternatives–running models with data in plaintext, or using data that is anonymized or tokenized–are sub-optimal. Data in plaintext is a major security and compliance risk, and data that is anonymized or tokenized reduces its value, producing results that are less accurate or biased. Every row of sensitive data scrubbed from the record denies the organization running their models the kind of rich, detailed information that would produce better results.
With Cape Privacy, optimized for use with the Snowflake data cloud, financial services organizations can easily upload client-side encrypted data and run powerful AI predictions directly in Snowflake without decrypting. Cape's platform provides best-in-class encryption, built with the strong security of AES, secure multi-party computation and secret sharing. Data remains encrypted throughout the entire system, even while in-memory and in-use. Your data and your keys are never in plaintext. Furthermore, Cape Privacy offers automatic data use confirmation reports within the Cape Privacy user interface, as well as simple model monitoring and versioning. Metrics and model performance are rendered in easy to read graphical representations with end-to-end transparency, analytical views, and historical job run and model performance views.
With Cape Privacy, financial services firms can finally put their high-value, sensitive financial data assets to work, unleashing the full potential of that data, even while remaining fully encrypted. To learn more about Cape Privacy and our platform, visit our website. Or contact Cape Privacy for more information.