With seasonality, changing interest rates and not to mention compliance changes and the global pandemic, mortgage lenders are dealing with constant change.

While mortgage interest rates dropped 15 points to 3.56% in March, applications skyrocketed the following week by 55.4%.

The Mortgage Bankers Association (MBA) projects mortgage originations to total about $2.61 trillion this year, a 20.3% gain from last year's volume. Refinance originations are expected to double earlier MBA projections, increasing 36.7% to around $1.23 trillion.

At the same time, the mortgage industry is dealing with record refinancing, there’s uncertainty about existing loans that may need short-term payment suspensions or be entirely in jeopardy. This massive influx of borrowers seeking to refinance or asking for a forbearance of their current loans means mortgage companies are staff strapped, in a time where more people must work virtually. Capacity is a real issue now – managing paperwork is getting more difficult. We have  reached the point where lenders need to raise rates, just to weed out inquires versus actual loan applicants.

Data Digitization Enables Mortgage Lenders to Move Faster

Rather than outsourcing work, turn away business, or solely rely on human labor, mortgage companies should take increasing advantage of technology to expand capacity. Cognitive Machine Reading (CMR) helps mortgage processing businesses take on more work, working faster and more accurately, so customers get their loans processed quicker.

According to consulting firm Cognizant, "Approximately 60-70% of tasks in mortgage processes across [the] value chain are prep tasks, are logical, replicable and are direct candidates for automation."

While some mortgage companies started to leverage automation to process loans faster and manage staff productivity, most firms are late adopters when it comes to automation technology. And while automation, like RPA, has helped with tasks like loan pricing and title searches; it’s really about the digitization of data in these processes that makes automation much more useful.

Lenders Benefit from Fewer Errors and Delays

When mortgage processing companies adopt automation, they relieve employees from doing what is considered value-added but repetitive work like manual data entry into loan origination systems. More importantly, automation eliminates the need for error-prone, “stare-and-compare” work that's common in mortgage processing.

Cognitive Machine Reading (CMR) helps mortgage processing businesses increase their profits in an environment in which loan processing costs are on the rise. MBA recently reported that the average profit for each loan originated in 2018 was $367, down from $711 the year prior.

Mortgage processors literally unpackage documents — like bank statements and W2s — required for loan applications. Some documents are already in electronic format. But many others are physical documents and need to be scanned and manually placed in folders. And while traditional capture solutions, like OCR, can extract data from structured documents, most of the information resides in semi-structured or unstructured formats. Leaving staff to manually enter information into systems, like LOS.

The risks of processors accidentally transposing numbers and introducing other mistakes as they populate systems with mortgage information are high. As a result, mortgage processing firms are losing valuable time.

Data digitization solutions, like CMR, prevent such errors from occurring. CMR uses inference to automatically understand the context and validate against other information. When there are inconsistencies, CMR can detect those problems immediately. That's faster than having someone fill out the information on Friday and no one catching errors until Wednesday. 

What Features of Data Digitization Should Mortgage Providers Look For?

Mortgage providers should look past traditional data capture solutions that only ingests and digitizes some of their data. Solutions like CMR are built on fractal science, meaning it understands the patterns of words, regardless of language, and can ingest any type of data — structured, semi-structured, or unstructured.

Fractal-based automation means the AI engine is trained once. That means mortgage lenders don't have to spend time and resources training AI on every variation. This shortens implementation time from months to weeks.

Another key feature mortgage providers should look for is the ability to recognize, classify and extract data from images such as photos. This is important because customers frequently take photos of their W2s, payroll, bank statements and IDs and submit them through an online portal or app for mortgage companies to process. Processing such unstructured data in an automated way allows for straight-through processing of loan applications and significantly cuts down  manual efforts.

CMR integrates easily with existing business platforms and provides transparency for business users to view extraction accuracy, through the user interface.

Avoiding Automation That Only Addresses a Part of the Issue 

What mortgage companies must avoid is selecting solutions that offer only incremental levels of automation. For example, solutions that are limited to templates and font libraries choke in situations in which fonts seem similar but are out of order and/or have multiple languages in the same document. That's problematic because in California, for example, you might see a mortgage application in a language other than English. Or you may see two languages — both English and Spanish — on the same document.

Mortgage processing organizations should steer clear of automation solutions that can’t address handwritten documents or do signature comparisons. This is key for the mortgage sector since signatures are an important part of mortgage documentation.

Solutions that cannot perform  any type of inferred information in multipoint lookup also should raise red flags because they lack the ability to do document comparison. And mortgage companies should avoid automation solutions that lack Machine Learning because they will not be able to improve over time as they get more experience processing data.

In Closing

Mortgage companies that adopt data digitization technology will free up valuable resources to better support customers throughout the mortgage loan process. They’ll be able to adjust to changes and take advantage of opportunities as they arise. And they’ll prevent errors and enable straight-through processing, which in turn, enables their operations and customers with optimal efficiency, performance and speed.

Many industries have taken advantage of the benefits of automation. The time is now for the mortgage industry to do the same.