Welcome to OpenForum. Join our global community: Sign up, Login or Learn more

Kreditech: Big Data Scoring for Consumer Lending

The Hamburg (Germany) based startup employs machine learning and big data to revolutionize the private banking sector.

Photo of Martin Schoell
0 0

Written by

It seems inevitable, but will in ultimately be for the consumers advantage? Kreditech is another startup, surfing the big data wave. Leveraging thousands of data points, Kreditech is predicting the creditworthiness of its customers and subsequently calculates loan rates and amounts - within a matter of seconds. The service offered by German fintech startup is a thorn in the flesh of privacy groups in Germany. But to be fair, the German market is generally more sceptical to every matter where consumer data is collected and stored by the government or private institutions.

Let’s take a look at their value creation first:

Kreditech Scoring Model

Traditional credit scoring is a balky process. It consumes a lot of time and resources and especially in developing economies, it is not available for every prospective customer. Kreditech offers an instantaneously available service - provided they are granted access to the customers private data, the more the better. Unlike traditional credit scoring services, Kreditech claims that their scoring is rather a prediction of future creditworthiness than an assessment of past behaviour. Your Amazon purchases, your location data, your Facebook profile and friends, your devices, your typing behavior, even the fonts installed on your computer will influence the rating you receive. Based on machine learning, the servers in Kreditech’s headquarters make smarter and smarter decisions every day. But the downside is: due to the phenomenal amount of data processed, the CEO admits that the system has already reached a level of complexity where for a particular decision, he is not able to elaborate on the concrete reasons for the decision of his system. Which raises the concern: if not even the employees understand the interaction of the 20.000+ data points considered, how can they be sure that their system really evaluates for causation, rather than just correlation?

Having pointed out these concerns, it is remarkable how much value the company is able to create. Aiming their service to developing economies and short-term loans, they manage to disperse the concerns - simply because their customers don’t have another chance to get a credit at all.

And how do they capture value?

Their value capture model is pretty straightforward: you get a credit, you pay your interest. What is notable though is, that since Kreditech is mainly offering short-term loans, the effective annual interest rates are not uncommonly in the range of 3000% to 6000%. I haven’t seen their P&L, but that should result in a pretty satisfactory profit margin!

Is their operating model sustainable?

Probably not as it is at the moment, but while they are able to reap the benefits of their underprivileged customers in developing economies, there seems to be no economic reason for them to alter their operating model. And Kreditech is also preparing for the time after their direct-to market era: partnering up with business customers, such as the very banks that Kreditech is disrupting at the moment, they will expand their reach to more and more customers. And they do a great job in convincing investors about their planned trajectory: having raised $40 million in their series B last year and aiming for an even bigger series C this month, they are among the highest valued startups in Europe.

Final statement

So sometime in the not to far future, your car dealership might offer you a financing for your new Beamer convertible based on a Kreditech score. If you want a low rate, just make sure you’ve spent a nice vacation on the British Virgin Islands before applying and bragged about it to your various HBS friends on facebook - and if you’re doing all that from your iPhone 6, chances are your rate will be as low as it can be!


Join the conversation: