Automated Identity Verification
Auto Insurance Provider
With over 170 million internet users, Indonesia has one of the largest online population bases. Still the existing practices of KYC in Indonesia include challenges in identification and verification of customers with manual process.
To achieve the government's vision of financial inclusion during the booming digital economy, all identity verification requests should be routed through the Indonesia NIK database. A low cost digital infrastructure to verify the identity of an individual is a necessity.
An ideal process for KYC verification should be real time, offer multimodal authentication options and should adhere to all compliance and data protection laws and practices.
Client is one of the pioneer automotive insurance providers in Indonesia, with over 4 decades of consistent best service delivering experience and an unlimited list of satisfied customers. Client wanted to solve the issue of growing inherent risks and inefficiencies in the existing KYC and customer on-boarding processes to achieve its financial inclusion goals.
The major challenge for automotive Insurance companies are:
The growing inherent risks and inefficiencies in the existing KYC verification and customer on-boarding processes.
The main reason for the risk and inefficiencies is their dependence on direct data input by customers and verifying the same same against the single source of eKYC truth manually.
This leaves room for manipulation, duplication, and poor quality of data/images.
To address these issues, auto Insurance technology companies had started adopting additional processes and exception handling techniques that in return had increase the operational costs. Given a difficult geographical landscape, implementation of automated process of KYC information extraction, verification and customer on-boarding, would be critical for Indonesia to achieve its financial inclusion goals keeping the operational costs low.
State-of-art Computer Vision, NLP (Natural Language processing), and OCR (optical character recognition) techniques, with text patterns extraction and business heuristics, are used to digitise the KYC information extraction and verification process.
In Indonesia the Insurance providers majorly receive the following KYC documents for verification KTP, STNK, and SIM. And NIK, Nama (name) and Berlaku Hingga (document expiry date) are the main KYC entities used for customer verification.
Custom computer vision model is trained for important text region detection in the KYC images, python based OCR used for image to text extraction, NLP named entity extraction technique used for extracting important entities from raw text.
Client had stated seeing a significant improvement in the overall KYC verification and on-boarding process.
PERFORMANCE IMPROVEMENT HIGHLIGHTS-
Decreasing the overall customer on-boarding costs and improved service delivery.
Achieved document verification within 1/10th of the time taken manually.
Achieved high accuracy, with near-zero margin of error.
Within 6 months, reduced manual efforts for KYC verification by 70%
Most of the repetitive document digitisation processes got automated under auto-insurance claims and registrations flow.