US 11,922,495 B1
Automatically determining adverse action reason codes
Juan Hernandez, Oakland, CA (US); Jianye Xu, Alameda, CA (US); Wensi Xu, San Francisco, CA (US); and Ihsan Kabir, Oakland, CA (US)
Assigned to BLOCK, INC., Oakland, CA (US)
Filed by Block, Inc., San Francisco, CA (US)
Filed on Dec. 21, 2021, as Appl. No. 17/557,806.
Int. Cl. G06Q 40/03 (2023.01); G06F 18/214 (2023.01); G06Q 20/32 (2012.01); G06Q 20/40 (2012.01)
CPC G06Q 40/03 (2023.01) [G06F 18/214 (2023.01); G06Q 20/3276 (2013.01); G06Q 20/4016 (2013.01); G06Q 20/407 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method implemented by one or more computing devices of a payment service, comprising:
determining, by a training component of the one or more computing devices, signals associated with previously denied lending queries, wherein each signal is associated with a value of a risk metric and an indication of whether the signal increases or decreases a corresponding risk metric;
training, by the training component and using the signals, a model to generate output signals for new lending queries, wherein each respective output signal indicates (i) that a new lending query is approved or (ii) that the new lending query is denied and a respective reason why the new lending query is denied;
in response to receiving, from a payment application executing on a user device of a user of the payment service, a lending query associated with the user, applying, at a first time and by a lending-decision component of the one or more computing devices, the model to the lending query to obtain an indication that the lending query is denied and a reason for declining the lending query, wherein the lending query and an indication of a decision to decline the lending query are provided to the training component for updating the model;
sending, by the lending-decision component, a notification on a user interface associated with the payment application that the lending query is denied and the reason to the payment application, wherein the notification includes an indication of one or more actionable recommendations for modifying a particular risk metric associated with a particular signal;
monitoring, by the lending-decision component, interaction data associated with the user, wherein the interaction data indicates one or more interactions with the payment service and via the payment application after the lending query is denied;
in response to receiving the interaction data, determining, at a second time and by the lending-decision component, that the user is approved for a loan from the payment service, wherein the determining is based at least in part on a modification to the particular risk metric, and wherein an indication of a decision to approve the loan is provided to the training component for updating the model; and
in response to receiving an indication that the user intends to apply the loan to a purchase for an item, issuing, by the lending-decision component, the loan to the user, wherein the user is permitted to use funds associated with the loan for purchasing the item and repay the loan in one or more installments.