AI Case Study

Funding Circle provides the only access to finance for 21% of its SME users SMEs by using machine learning in loan underwriting

Funding Circle enables small and medium businesses to apply for business to business loans through its streamlined application process, which includes machine learning for part of the underwriting process.

Industry

Financial Services

Banking

Project Overview

According to City A.M., Funding Circle "uses the traditional underwriting process used by the banks to ensure a pair of human eyes looks over each loan, it combines this with machine learning. 'There are different complexities in business lending; when we launched, we didn’t have any data, so we employed experienced credit underwriters to look at every single loan – regardless of how big or small the loan was. And as you build up more data, you can vary the level of experience you need for different loan sizes, and combine the use of algorithms and artificial intelligence.' While the banking world will often price loans based on their size, Funding Circle aims to ensure that it’s the level of risk that’s the focus, rather than the amount of money being borrowed."

Reported Results

Funding Circle contends:

* 21% of borrowers indicated they would not have been able to obtain alternative finance without Funding Circle
* 61% of borrowers claim revenue increases resulting from obtaining loans through Funding Circle
* 47% of borrowers stated an increase in profits

Technology

Function

Finance

General Finance

Background

From City A.M.:“They might be the most unloved part of the banking world, but SMEs contribute to most of our GDP. This is about a fundamental re-wiring of an industry, which Funding Circle hopes to drive... banks will continue to retreat from the SME space as the impact of the EU referendum sinks in. In fact, disconcerting figures from the Bank of England indicate that net lending to SMEs has been shrinking since the summer of 2016."


Benefits

Data

Specifics not disclosed; historic loan repayment/default data likely