Automated Document Processing for India’s Leading Insurance Company

6 min readJun 23, 2021

Challenge

The commercial insurance industry operates using document-intensive processes. These documents, such as policy forms, loan applications, etc. which are mostly handwritten, contain data like applicant names, place, phone number, or patient’s health history, which is essential to business processes.

All of this data needs to be extracted from digital documents or jpg files to perform tasks like process loan applications, analyze customer sentiment, issue policies, etc. Today, many companies manually extract data from scanned documents like PDFs, images, tables, and forms, or use simple handwriting recognition tools that require manual configuration which oftentimes requires reconfiguration when the form changes. All this is time-consuming, error-prone, expensive, and difficult to scale.

Confronted with a similar problem, one of India’s Leading Insurance Companies reached out to us at Applied Cloud Computing. Out of the nine types of insurance policies that they offered, they received over 700 to 800 policy proposal alone for health and long-term home insurances each day. Their insurance agents took pictures of the filled forms and sent them over to the office. The employees then read data from the images and filled excel sheets under different fields — like name, date of birth, Adhar card, etc.

The proposal forms had a separate format for each insurance type and most of them had separate boxes for capturing the data. The majority of these forms were handwritten. Therefore, diversity in human writing types, spacing differences, and irregularities of handwriting resulted in less accurate character recognition. A variety of image qualities submitted by the insurance agents added to the problem. These images generally featured some form of the background image which made it reading the text more difficult and increased the processing time.

To overcome these challenges, the Insurance company wanted ACC to build a solution that accurately reads data from the images and creates excel files with proper fields so that they could use the structured data efficiently to issue policies and become a valuable differentiator in customer experience and operational efficiency.

Solution by ACC

We, at Applied Cloud Computing, proposed an intelligent image/document processing solution using Amazon Web Services.

We used Optical Character Recognition (OCR) to read and process the images, accurately extracting text, handwriting, tables, and other data without any manual effort. It then classifies and labels the unstructured content in documents and pull-out insurance-specific entities, creating excel sheets with proper mapping.

Despite the variety in writing styles, scribbled characters, unaligned text, spacing differences, colored backgrounds and subpar quality of images, we were able to achieve 70–80% accuracy for the company in processing the images and reduce human efforts by 50%. Once the excel sheet is prepared, employees need to go through the converted data fields to ensure 100% accuracy.

With the proposed solution the Insurance company was able to quickly automate document processing and take action on the information extracted to issue policies.

Key Features:

Fast: Works automatically and eliminates the need to write code for data extraction — meaning you can extract the details you need quickly.
Flexible: Extracts text and data from virtually any type of document — including forms, images and tables.
Accurate: Uses machine learning to extract data — reducing manual human error.

Benefits

Faster data processing — Extracts and digitizes data no matter the document type to quickly process thousands of forms each day.

Higher accuracy of data — By using the proposed automated solution you can process documents faster and more accurately, reducing errors caused by manual entry.

Improved employee productivity — The proposed solution removes the manual process of pulling out entries from documents and entering information into various systems, enabling your employees to spend more time on value-adding business tasks.

Cost savings — Automating document workflows reduce the complexity of data extraction and analysis, reducing the average cost per document. Additionally, it is based on a pay-as-you-go pricing model which further optimizes your cost.

Secure — The proposed solution conforms to the Amazon Web Services (AWS) shared responsibility model, which includes regulations and guidelines for data protection. AWS is responsible for protecting the global infrastructure that runs all the AWS services; therefore, you need not worry about data being leaked or used by any others.

Higher Customer Satisfaction — Improve customer satisfaction by providing your clients with more accurate information faster and more efficiently.

Future Scope

Create smart search indexes

Your organization may have thousands, if not millions, of documents, records, and reports stored in archives. Finding what you need in such volume can be an arduous, manual process. By using our solution, you can create a digital smart index of documents, searchable by key terms or key-value pairs.

Sample business applications:

In finance, banks or insurers can create smart search indexes to quickly find a specific customer name, or locate a record with specific parameters (such as all loans with x% rate).
In healthcare, parties managing a clinical trial can use our solution to gather the text from documents, understand that text, and then build a searchable database — which can be used to identify patients that fit trial criteria.
In government or the public sector, institutions can create indexes to easily find reports and research on a specific public policy.

2. Build automated document workflows

Across industries, our solution can be used to automate common processes. For example, financial institutions and lenders can automate loan applications by using the information contained in documents to initiate all of the necessary background and credit checks to approve the loan — so that customers can get instant results of their application rather than having to wait several days for manual review and validation. Likewise, insurers can rapidly process claim and forms; and government agencies can automate applications such as driver’s licenses and/or identification cards.

3. Analyze trends in data over time

Today, businesses get smarter and more efficient by understanding patterns and trends across their company over time. A lot of that valuable data is locked in physical documents containing forms or tables. Traditional methods can only pull-out text, making it hard to understand the structure and logic of the data. Our solution automatically identifies form, labels, and values and extracts information from tables without compromising the structure, so you can understand patterns over time across millions of documents.