In 2020, the Bounce Back Loan Scheme (BBLS) saw UK big banks receive approximately 100,000 applications in one day. Barclays announced they would start issuing payments within 24 hours of receiving the applications – an unprecedentedly fast turnaround.
How did these banks manage the surge in loan applications in such a timely manner? Part of their strategy involved deploying an automated credit decisioning system, such as Appian, to assess the creditworthiness of thousands of small businesses.
An automated credit decisioning system is a program that chooses whether to approve or deny a person or business’s loan application. The automated decisioning engine (ADE) is essential to training this system. The ADE is trained on high-quality, complete loan application data. The result is an automated credit decisioning system that should produce loan decisions within seconds, not minutes or hours.
For use cases like the BBLS, it’s essential for both loan recipients and lenders that the decisioning system vets the loan application data quickly and carefully. The scheme offered “loans, not grants” (as clarified by Stephen Jones, chief executive of UK Finance), so borrowers had to be able to start repaying the loan in 12 months. Consequently, lenders had to handle loan applications efficiently to ensure timely disbursement while maintaining strict due diligence.
Automated credit decisioning systems must be accurate, fast and user-friendly. Hence, this article will examine three use cases for high-functioning automated credit decisioning systems.
Small businesses, in particular, benefit from the swiftness of automated credit decisioning. Why is this?
They generally require speedy access to funds. Since small businesses represent 99.2% of the UK private sector, it might benefit lenders to whittle decision-making down to seconds or hours rather than days or weeks).
Though this might seem like a pipe dream, credit decisioning engines like Planky or Creditsafe promise credit decisions within a few seconds.
Small businesses also require consumer trust, which means building a system that delivers accurate outcomes – even if the system is handling 100,000 other applications simultaneously.
However, small businesses may not be the only ones that benefit from accurate (and timely) automated decisioning. Underrepresented groups – such as Black, Asian and Minority Ethnic (BAME) – have traditionally been denied credit, excluding them from financial services.
Automated credit decisioning systems can reinforce exclusion via rigid, rule-based systems. Or, in contrast, these systems can open up access to financial services by reviewing alternative data sources. Let’s explore this further.
Collating typical consumer data –such as key financial and personal data points and trends –and using it to train systems helps develop high-performing automated credit decisioning models. Yet inflexible, automated decision-making might disadvantage underrepresented groups by reinforcing biases or overlooking unique circumstances.
Case in point: Groups without access to traditional financial resources – such as bank accounts, credit history, employment incomes, etc. – will not pass creditworthiness assessments, denying them access to personal and business loans.
ADEs using prescriptive rule-based systems will automatically deny credit applications lacking verifiable financial data in the form of required documentation (e.g. bank statements, credit history, etc.). But consider a quote by Canadian journalist Malcolm Gladwell: ‘The key to good decision-making is not knowledge. It is understanding. We are swimming in the former. We are desperately lacking in the latter’.
It may be a fanciful quote, but it is surprisingly easy to collect and maintain data without understanding what it really means (as M&A firm Langcliffe International found). Serving underrepresented groups means understanding that the definition of creditworthiness might extend beyond a credit score or individuals with a healthy income-to-expense ratio.
Lenders increasingly understand that alternative data sources, such as rental payment history, subscription services, mobile phone contracts (and more), can also demonstrate a borrower’s creditworthiness.
%20(1).png)
Various firms are working to automate outside of traditional financial standards, including:
These case studies show us that lenders don’t need to operate traditional credit scoring models if they (really) understand borrower data and how it can be systemised (even from alternative data sources like a borrower’s mobile phone contract).
Providing credit to underrepresented and credit-invisible borrower groups represents a socially and financially valuable project (e.g. fintech, Payaga states that their alternative data product converts an average of 30% more customers for lenders). We hope to see more groups accessing credit in 2025.
You might also note that these alternative credit decision systems are built via a fintech partnership, representing a valuable use case beyond the borrower/loan applicant.
Automated credit decisioning systems can have value for stakeholders other than borrowers. By partnering with other fintechs to fill the technological gaps, fintechs can build products that banks and other loan providers can adopt.
Let’s examine a case study where a fintech partnered with a funding provider.
YouLend is an embedded finance provider, enabling various customers in fintech – banks, paytechs, marketplaces, etc. – to provide flexible funding.
YouLend has facilitated over 150,000 funding instances, partially powered by Evolution AI’s automated data extraction technology. From 2022 onwards, YouLend has worked with Evolution AI to build an automated credit decisioning system.
Jessica O’Hare, Head of Product at YouLend, elaborated on their deployment of automated data extraction: “So we use AI to extract information from bank statements in our credit decisioning model, not to make the credit decision necessarily, but to get the information out of the bank statements into our platform, so our platform can make an automated decision. This obviously reduces the manual effort required but also massively speeds up the time-to-decision and the time to fund.”
YouLend’s collaboration reflects the importance of human-technology and technology-technology collaborations. ‘By 2025, smart workflows and seamless interactions among humans and machines will likely be as standard as the corporate balance sheet,’ McKinsey enthusiastically suggests. After all, technological partnerships can create socially important technologies like automated credit decisioning systems.
Do automated credit decisioning systems work? Yes, these systems can deliver accurate decisions based on the available data. A better version of the question might be, ‘Who do automated credit decisioning systems work for?’.
Three demographics in particular stand to benefit from automated credit decisioning systems: small businesses, underrepresented groups and other fintechs.
When made from best-in-class technologies, automated credit decisioning systems will deliver fast and accurate credit decisions to a rapidly widening pool of borrowers in an uncertain economy.
Follow Evolution AI on LinkedIn to stay updated with our latest insights.
