Summary:

ZestFinance, the machine learning meets personal loans startup from former Google CIO Douglas Merrill, has raised a $20 million series C round. The company’s model analyzes more than 70,000 variables in trying to provide good loans to folks with bad, or no, credit.

Structure Data 2013 Douglas Merrill ZestFinance
photo: Albert Chau

ZestFinance, the Los Angeles-based startup using machine learning to assess credit risk for personal loans, has raised a $20 million series round of venture capital. Peter Thiel led the round, with Northgate Capital, Matrix Partners, Kensington Capital Holdings, Eastward Capital Partners and Lightspeed Venture Partners also participating.

ZestFinance is the brainchild of former Google CIO Douglas Merrill (pictured above at Structure: Data 2013) and Shawn Budde, the former Head of Subprime Credit Cards at Capital One. The company provides an underwriting model to lenders that aims to serve individuals with bad credit or who are otherwise underserved by traditional banking products.

The ZestFinance model, simplified

The ZestFinance model, simplified

The magic behind ZestFinance’s methods lies in the approximately 70,000 variables its models considers, which are then analyzed using a number of machine learning algorithms. Once the algorithms have done their job, humans step in to apply some logic, judgment and  context to the results. All told, ZestFinance claims a roughly 60 percent improvement over traditional underwriting models, and repayment rates 90 percent higher than traditional methods.

In January 2012, ZestFinance raised a $73 million series B round that included $23 million in equity financing and $50 million in debt financing.

Although ZestFinance was early to the game of serving the banking industry’s underserved, it’s not alone. ZebitAvantCredit and Kreditech are two out of a handful of companies worldwide providing similar methods, albeit with varying business models. In the small business world, companies like Kabbage and Capital Access Network are looking at alternative types of data to provide financing.

Comments have been disabled for this post