Using a unique combination of AI and data science to create our latest tool
Today, we launched Borrowing Power, a new tool that combines AI, data science, and innovative technology to allow customers to borrow at the best possible rate.
Users are shown what makes up their ‘borrowing power’ within Zopa’s app along with bespoke actions that can help them to improve it. Uniquely – the score is directly linked to the Zopa loan it unlocks, so that a customer can see immediately if they are eligible and at what rate. They can also see the impact on loan availability and price if they improve their score by a set amount – linking improvement to a tangible outcome.
We’ve used cutting-edge technology that operates on an advanced non-linear credit risk assessment model, integrating machine learning and AI to create better outcomes for our customers. Customers are given a bespoke ‘borrowing power’ score, which is a simple rating between 1 and 10 (rather than 0 – 999). It’s made up of five components, including a combination of credit rating data, credit utilisation, credit limits, hard searches and affordability based on personal circumstances.
Historically, credit reference agencies have used machine learning to determine the credit risk of a customer, or in more simple terms – used AI to predict if customers will default or not. Typically, reference agencies have used a linear model, which means they have only been able to provide customers with generic guidance on which variable will help them improve their credit score the most.
Our unique approach uses powerful machine learning techniques to understand what is happening in the black box of the credit referencing agencies. We then interpret the data, along with additional information provided by the customer to examine whether they can improve an aspect of their credit profile, and in what time period. In some cases, an improvement could be seen in as little as two months.
Zopa has tested its combination of technology and machine learning and found that it can lower costs of borrowing for customers because it can accurately link changes in their borrowing power scores back to its own loan products, with increases in borrowing power resulting in a cheaper cost of credit.
We’re uniquely placed to be able to deliver this capability as we build on our 14 years’ of lending data, coupled with information from traditional credit bureaux and a combination of machine learning, and tried and tested technology.
Didier Baclin, Zopa’s Chief Product Officer, said: “Customers deserve to know their eligibility for credit, current credit scoring is merely scratching the surface. We have effectively broken open the black box to understand what is going on and, combining this data with additional information about the customer, are able to give bespoke actionable insights to our customers that could enable them to improve their credit risk in a short time frame and then ultimately be able to borrow from Zopa at a better rate.”
In all cases, we show real rates so that customers know exactly what their rate of interest will be should they decide to take up a Zopa loan through the new app. Zopa also only performs soft searches on customer’s credit history prior to application so borrowing power can be used over time with no impact on their credit score, until a loan is taken out.
Sarah’s Real Money Story: a tale of two relationships
Zopa customer Sarah shares what she’s learned from the two very different relationships that have shaped her Real Money Story