Expanding Credit Access to the Most Underserved
Traditional underwriting with respect to reaching the underserved is flawed in a few important ways.
-
You need credit history to get credit
-
Lenders often depend on limited and unverified data sources
-
Most modeling approaches are either wholly subjective or have inferior predictive power
This resulting lack of precision is compensated for by declining many and approving few, perpetuating exclusion. This is especially pronounced for people of color, minorities, women and other marginalized groups.
Uplinq has incredible potential to change the way small business credit decisions are made by unlocking access to richer insights & creating a workflow that fosters greater transparency.
Archie Puri
Former CPO, Galileo & GM, Braintree
Archie Puri
AI Isn’t Always the Answer
Below are some common challenges faced when developing risk models.
-
1. Expert Models
Even when they are effective, expert models are individual assets rather than business assets, limiting their scalability. -
2. Decision Trees & Sequential Filtering Methods
While these may perform adequately for the least risky, most conventional, or historically most represented segments, they are ineffective at balancing trade-offs. -
3. Advanced Methods
Techniques like multivariate regression can be extremely powerful, but many modelers struggle with feature extraction, common data missingness, collinearity, and overfitting. The same issues apply to AI/ML, compounded by challenges in reviewability and gaining insights from the models.
It’s reasonable to assume that common barriers to inclusion stem from a lack of product and marketing fitment, as well as an inability to understand information and behavior patterns that predict credit risk before credit behavior has been demonstrated.
To address this, we need static, observable models. AI faces significant challenges in this regard.
To address this, we need static, observable models. AI faces significant challenges in this regard.
Over a Trillion Served
Uplinq has successfully addressed all these challenges, facilitating over a Trillion dollars in capital for SME business owners worldwide.
By utilizing a wide array of data sources beyond those reliant on credit history we’re able to gain a comprehensive view of the customer, with models trained to improve predictability for traditionally underserved segments.
By utilizing a wide array of data sources beyond those reliant on credit history we’re able to gain a comprehensive view of the customer, with models trained to improve predictability for traditionally underserved segments.
