loading

AI in Mortgage Lending Automation

Over the last two decades, mortgage lenders have faced both challenges and opportunities in a highly regulated yet growing market. While lending in banks has bounced back after the 2008 global financial crisis with improved risk management, the non-bank lenders and fintech firms have gained market share in this space. During the COVID pandemic, demand for new mortgages and refinances surged, and the remote working model required lenders to facilitate digital mortgage origination.

The demand for mortgages is growing despite the rising mortgage rates. According to the Mortgage Bankers Association (MBA), mortgage origination will be upwards of $2.5 trillion each year over the next three years. Freddie Mac recently published a forecast indicating that despite higher rates and slowing price increases, the housing market will stay steady. From 7.1 million in 2021 to 6.9 million in 2022, home sales are expected to rise to 7.0 million in 2023. With this growing trend in the industry, mortgage players will have to stay competitive to acquire and retain customers while optimizing the processes.

Why digital transformation?

After the subprime mortgage crisis, traditional lenders are now focused on driving growth without compromising on the risk and quality of loans while adhering to the regulatory requirements. 

Over the past decade, consumers’ expectations from their financial service providers have significantly changed with the shifting demographics and evolving digital technologies. The mortgage market has seen an influx of digital native non-bank and fintech lenders gaining market share from traditional lenders. In the US, more than 60% of mortgages originated from non-bank lenders and have grown beyond double-digits. 

The need for mortgage lenders to provide a superior customer experience is no longer a choice but a mandate. A focused digital transformation is vital to understand and effectively engage with their customers and enhance their lifetime value.

From a financial perspective, mortgage continues to be a slim margin business, particularly in the sustained low-interest-rate environment. Our analysis of a sample of mortgage lenders indicates that net margins are in the low single digits (between 1-3%) since many lenders need to stay competitive on mortgage rates while investing in their operating capabilities. A large number of lenders are even running at a net loss ever since the pandemic-led growth has tapered. Therefore, leveraging technology to automate business processes is the only lever that is available to these lenders to reduce operating costs and boost margins. 

Lenders need to capture the opportunities of growing demand with inherent risk management and to do so, they need to improve their customer targeting and retention strategy through hyper-personalized offerings. Simultaneously, lenders will need to leverage digital technologies to fast track their processes and drive operating efficiencies to remain competitive in the industry. 

What is digital transformation in mortgage lending?

To digitally transform the mortgage processes, lenders need to take a front-to-back assessment of their processes to enhance customer experience and lifetime value. A typical mortgage process covers the following steps:  

Marketing for Customer Acquisition – The traditional analytical approach falls short in providing a comprehensive view of customers and results in poor returns on marketing spending. It is becoming increasingly important for lenders to understand their customers better, including their pivotal life events, drivers, and motivations to purchase mortgages and other financial products. Additionally, understanding borrowers’ journeys will help lenders provide a hyper-personalized experience to increase their confidence in purchasing.

Origination (including Document Submission and Reviews) – To ensure quality borrowers, mortgage lenders review documentation for identification and income. Critical data from these documents are entered into mortgage systems. The manual process is slow and prone to errors, requiring underwriters to spend more time ensuring that the information provided is accurate. 

Underwriting – The most critical step of the mortgage process, underwriters tend to spend more time and effort to organize missing or incorrect information collected during the preceding step. Although the actual task of underwriting might not undergo drastic changes, digital transformation throughout the process can reduce the workload on underwriters and simplify their jobs while allowing lenders to appropriately assess risks. 

Customer Service (including Engagement, Retention, Refinance) – Until a few years ago, a lender would be less worried about their customers refinancing mortgages through another competitor. However, with the refinance boom during the pandemic, maintaining their loan portfolio has been one of the biggest challenges for lenders. Therefore, digitally-enabled customer services and proactive retention initiatives are the drivers for enhancing customer lifetime value. 

Role of Data and AI in Lending Transformation

The digital transformation journey for lenders starts with understanding customers better using a Customer Data Platform and associated descriptive and prescriptive analytics. The ever-increasing footprint of real-time data makes it essential to use modern-day platforms that can support real-time data capture and processing. A customer data platform provides organizations with a unified platform to capture and process data across all customer touchpoints.
Leveraging AI to automate document processing is a huge step toward achieving operational efficiencies. Loan application processes are document-intensive and time-consuming with applications to be examined, scrutinized, and verified. The use of AI/ML technologies reduces the cycle time for the document verification processes including, but not limited to, automated classification, intelligent extraction, discerning insights across multiple documents and applications, and improving overall operational efficiency. This drives faster processing of the loan applications and expedited loan disbursements resulting in gaining a competitive advantage.

Challenges for Digital Transformation

Lenders face the following challenges in front-to-back digital transformation:

Legacy Processes and Technologies – Many traditional banks and lenders have been operating with processes that were designed for decades-old customer expectations. As customers’ expectations have changed over the years, institutions have modified some processes and added technologies to drive automation, only to increase the number of bottlenecks. To compete in the changing industry landscape, financial services firms need to reimagine their processes keeping data and AI at the center.  

Organizational Initiatives and Adoption – The biggest challenge in a mortgage process across various functions in a financial institution is identifying the starting point in the process. Often the resistance from various teams in taking initiatives to adopt technologies makes it difficult to transform the process. These could be further compounded by certain questions such as Do we take the top-down or bottom-up approach? Should the starting place be automating document processing or developing marketing analytics? These are important questions and each organization needs to find an approach that works for them but it is critical to identify the starting point and innovate with a fail-fast approach rather than not starting the transformation journey at all. 

Regulatory Considerations – Post-financial crisis regulations have created several challenges for traditional banks. Consumer lending continues to be a regulated landscape and applying AI without considering the laws and ethics of lending processes will only amplify challenges for the financial institutions or lenders. There are areas where human interventions are required and a comprehensive digital transformation considers these compliance guidelines. 

Quantiphi’s Recommended Approach  

Quantiphi helps banks and lenders with end-to-end transformation by developing target states with automated processes and digital technologies and implementing solutions to deliver tangible value. We recommend our clients adopt a comprehensive view of the transformation and take the first step towards execution to achieve quick wins. 

Our proven 4A (Assess, Assimilate, Align, and Act) approach helps us empathize with our clients’ challenges and allows us to identify the opportunities and co-create a roadmap to realize the full potential of AI-led digital engineering use cases for mortgage transformation. This approach helps our clients identify quick wins and define return on investments to progressively build solutions leading to end-to-end digital transformation. 

Our Customer Data Platform (CDP) solution, OneCustomer, enables banks and mortgage lenders to collect and consolidate data from multiple sources (structured, unstructured, real-time, etc), resolve identities, and create a single 360° view of the customers across multiple touchpoints. OneCustomer is powered by built-in intelligence with pre-built models to predict purchase propensity, churn, lifetime value (LTV), sentiment, etc., enabling banks and lenders to improve conversion, retention, and customer experience. 

A key automation area in mortgage lending is processing documents using AI and removing manual intervention to the extent possible without compromising the compliance guidelines. Our document automation solution simplifies cumbersome mortgage processes with reduced human intervention, especially in repetitive tasks. The technological solutions include platform modernization, workflow management, document classification and extraction, income and asset verification, employment verifi­cation, title verification, appraisal management, automated compliance, and decision-making. These technologies have led to enhanced operational efficiencies, reduced costs, and shrunken turnaround times for the mortgage processes.

Along with technology, people and processes play a critical role in the success of digital transformation initiatives. We help customers drive this change using our Transformation Management Services, which cover roles and responsibilities, job descriptions, process maps, value stream mapping, Standard Operating Procedures (SOPs), training charters, and communication plans. We ensure that the people and process transformations complement the technology initiatives and help clients adapt to the change. 

Quantiphi’s Success Story

Quantiphi has helped both banks and non-bank lenders in end-to-end lending modernization. Typically, a customer starts with our advisory engagement to reimagine the lending processes using AI followed by developing a proof of concept (POC) for a specific use case such as customer acquisition analytics or document automation. Once the value of a use case is delivered, the solution is scaled with workflow automation. 

The following are the success stories with clients delivering specific use cases to transform mortgage lending:

Case Study – OneCustomer 

A large financial institution wanted to address the challenge of customer persistency and implement multiple customer analytics use cases such as purchase propensity and customer segmentation. The firm also wanted to develop a recommendation engine for the next best action. Quantiphi used its OneCustomer solutions and model accelerators to address the challenges. Using the client’s data enriched with relevant 3rd party data, Quantiphi created a comprehensive 360-degree view of the customer that gave front office and customer relationship managers a real-time view of customer data and AI-powered predictive analytics to drive customer acquisition and retention. This solution provided the client with a 27% increase in purchase propensity for their mortgage product among other crucial success factors.

Case Study – Automating Document Processing

A large mortgage lender faced the challenge of processing a huge volume of income verification documents ranging from standardized templated documents like passports and driver’s licenses to non-templated pay stubs and employment verifications. Quantiphi developed solutions to save time and effort for their underwriters by leveraging powerful parsers that automated the entire workflow – from ingesting documents to successfully classifying 330+ document types with an accuracy of over 99% on 79M pages. The entities were auto-extracted from 32+ documents with a field-level accuracy of over 80%. This workflow was then seamlessly integrated into their post-processing and downstream applications and has reduced the time to process loan applications from 2-3 days per application to a mere 7 minutes.

Get in touch with our experts to modernize your lending ecosystem.

Written bySanjeev Sethi

Get your digital transformation started

Let's Talk