Small but growing market faces risk
In the massive American automotive market, “buy-here, pay-here” dealerships are a small but rapidly growing portion of the market. These dealerships offer on-site auto loans to buyers who may have no credit whatsoever, and even no bank account. This process allows disadvantaged customers access to the cars, trucks, and vans that they need, but it also presents significant risk to the dealership. With no credit to check and no verification of employment, account balance, or buying history, loans can be at significant risk of default.
As the buy-here, pay-here market grows, dealerships are accumulating lots of data on their customers outside of the traditional data that banks store, like credit history.
AI offers one dealership a way forward
A Texas buy-here, pay-here chain decided to look into AI solutions that might enable them to better assess client risk, optimize individual interest rates, and forecast inventory needs. With a growing number of locations in the state, the dealership had ample but unstructured stores of data on their clients, as well as many years of inventory data across different cities and regions. Despite their access to information, the dealership was not able to leverage it to make good predictions about client risk that would help them determine the best interest rates to charge particular types of customers and thereby lower their own risk in lending. Their data did not include traditional auto loan information like credit scores due to their target market.
The dealership hoped to find a long-term strategic AI partner that could help them extract value from their data and make their business more profitable by increasing their certainty in lending.
The company selected CrowdANALYTIX as their AI partner in part because of the crowdsourced approach that CrowdANALYTIX offers: clients need not worry about selecting the wrong data scientist for the job, because they have access to thousands of data scientists, whose joint efforts drastically increase the likelihood of success.
CrowdANALYTIX began by using the dealership’s large stores of data to build and establish a working credit scoring model that relied on the client data present rather than on traditional credit scores. Leveraging this model, CrowdANALYTIX then modeled different payment scenarios for different custom credit scores, so that the dealership could determine the optimal interest rate for varying types of customers. The business would be able to arrange the best combination of initial down payment, interest rate, and term of contract for each customer to maximize profit and reduce potential risk. Finally, CrowdANALYTIX used backlogs of historical sales data to make forecasts of how many cars the dealership would need to stock across locations and time periods. They also determined the factors that most significantly impacted the forecasts, enabling the business to make firmer, more accurate plans for the future.
All the models developed were then deployed on cloud servers and integrated to the company’s existing dealership dashboards and applications.
The buy-here pay-here dealership chain was able to reduce risk, increase profitability, and maintain optimal inventory, reducing unnecessary expenditure and risk of customer default. Early in the AI deployment process, the company has already reduced the risk premium on their loans by 4%, and expect to continue this increase as other components are added to their AI solutions.