Origination has become one of the mortgage industry’s largest expenses. The average cost to process a loan rose to $8,611 in fourth-quarter 2018 — where historically, from 2008 to 2018, loan production expenses have averaged about $6,200 per loan, according to the Mortgage Bankers Association (MBA). In addition, independent mortgage banks reported a net loss of $200 for each loan originated in the past fourth quarter, MBA reported.
Lenders are facing increased competition in a time-sensitive environment. One only needs to understand the success of products such as Quicken’s Rocket Mortgage to know that lenders, in order to compete, need to process mortgages faster than ever before. Needless to say, it is not an easy time to be in origination. Despite the challenges, however, there has never been greater access to innovation than right now.
It’s clear that the industry has started to respond to this challenge by introducing solutions in areas like computer vision, blockchain, artificial intelligence (AI), and robotic process automation. AI will play a major role in helping to reduce costs and improve productivity. Here’s how to leverage it effectively.
Understand that AI is well-suited at performing processes that can be converted to algorithms. In essence, the machine or bot is mimicking human behavior as a set of rules in the mortgage space. Some examples include income verification, pre-underwriting and structured data assembly (e.g., stacking order) within a loan packet.
Prior to engaging with any AI company, it’s best to document your organization’s pain points and workflow bottlenecks, and then determine the best outcomes your business is seeking. This way, you’ll have a benchmark to understand what improvements were made to your existing processes.
Your company may be looking to decrease the time it takes to process a loan packet or shorten the amount of manual review required. Either way, it’s important to define and document the current workflow prior to seeking an AI solution.
Proven track record
There is no shortage of AI solutions on the market today. Not all AI solutions are created equal, however. Much of the marketplace sells its technology with the expectation that clients will spend a massive amount of money on professional services and configuration.
For many of today’s lenders, this business model is insufficient. Many originators who purchase a horizontal process automation solution struggle to get the software accuracy rates high enough. If system accuracy rates are not high enough, then human validation is required and much of the automation benefits are lost. Lenders and originators can stay ahead of this problem by seeking solutions that have a proven track record for automation in the mortgage industry. Ideally, you want your vendor to show that their solution can mostly automate the mortgage life cycle without any human involvement.
In your proof-of-concept stage, consider testing the documents and related processes that your mortgage brokers or loan officers work with the most, so you can see if the software can work its way up to a performance level for data classification, splitting, extraction and stacking that would sufficiently reduce the level of manual labor done by a loan officer. If so, you’ll be making a much safer investment and are more likely to see significant return on investment (ROI) in the near term.
Depending on how your business operates, implementing an AI solution will either be a seamless experience or a big wake-up call.
AI cannot improve ineffective management or create consensus around a nonharmonious process. In fact, implementing an AI solution in this type of environment will be extremely challenging, if not impossible. Originators need to view these problems as an opportunity, rather than a threat. Although having differing processes doesn’t mean organizations can’t gain ROI from an AI platform, achieving it this way will be much more difficult.
As you implement your solution, try to seek a convergence of opinion among key participants. By following these steps to standardize the process, you won’t only reap the full benefits of your AI software, but also enable your workforce to establish more consistent practices.
AI is not a magic bullet and it will not solve every problem. AI performs best in discrete spaces where a bot can choose among classes or values.
With enough machine learning, it can determine how to accurately classify documents or the contents in each data field, or determine a loan’s risk. Since there are only a finite number of solutions possible, the AI software can learn within a discrete space how to get the right answer.
After these problems are solved, humans are required to re-enter the picture. A basic AI solution has difficulty performing tasks in which creativity is involved. Although bots certainly are smart, they can stumble without a set of rules in place. Setting the proper expectations for which tasks are ideally suited for an AI system will ensure that the initial deployment is successful.
When an effective AI system has been implemented, employees are able to focus attention on more meaningful tasks. Start looking at higher-level projects that have been put on the backburner as AI will free your employees from repetitive tasks.
Unfortunately, there has been a tremendous amount of misinformation regarding the impact AI will have in the mortgage industry. Originators and underwriters may hear that their jobs are being automated and, consequently, they think they’ll be out of a job.
But that doesn’t have to be the case. In fact, it’s more likely that your employees will be more effective and spend more time handling the more creative tasks that a bot doesn’t adequately perform well.
When intelligent machines are introduced into an organization, they will likely start performing the most redundant and mundane work that your employees want to hand off anyway. Effective change management will be needed when implementing any automation solution, to avoid employee resistance and apprehension.
• • •
Any significant technology changes, such as AI, involve significant barriers to entry. By implementing these practices into your organization, you can help ensure a reasonably seamless transformation to your AI solution. As a result, your organization will hopefully experience reduced origination costs and possess the ability to compete with the best in the industry.