How Mortgage Servicers Can Harness Generative AI to Strengthen Borrower Relationships

The correspondence, notices, and disclosures that mortgage servicers send to borrowers form a key part of the borrower relationship.

The language that servicers use within these communications has a powerful impact on how these messages are interpreted, the perception of the servicer, and whether borrowers follow through with an appropriate or desired response. However, too often these communications are rife with industry-specific jargon, legalese, and uncompassionate language. Servicers often recognize these issues but lack the time, resources, and technology to address them across their extensive libraries of communications.

The speed of text-based generative AI, such as OpenAI’s ChatGPT and Google’s Bard, can lend to the content optimization process, and these tools could be the solution servicers need. The challenge is deploying these technologies in a way that produces high-quality results and ensures safety and low risk while driving efficiency. In this blog, we’ll explore why better, clearer, and more consistent content should be a high priority for servicers, and the effective strategies for using generative AI to help.

The need for clearer, more consistent borrower communications.

Most mortgage servicers have prioritized compliance over clarity in their borrower communications, letting legal teams control final copy. As a result, communications typically contain complex phrasing, unfamiliar terms, and unsympathetic language that may be isolating for the average borrower. This presents an issue for borrowers, particularly when they need help, as 54% of adults in the U.S. read below a grade six level. During default and loss mitigation, the volume and complexity of communications increase, and it’s critical that borrowers understand the options available to them to take the appropriate steps to avoid catastrophe.

In July 2023, J.D. Power released its annual “Mortgage Servicer Satisfaction Study” which provides detailed information and insights on borrower experiences provided by more than 30 of the largest servicers in the U.S. The study found that the percentage of financially unhealthy mortgage customers rose to 54% from 48% during the past year and that default risk is also up 4% this year. It also found that the overall satisfaction among financially unhealthy customers is 107 points lower than among customers in the financially healthy category. Bruce Gehrke, senior director of lending intelligence at J.D. Power, states that this is a big problem for servicers: “When customers lack trust in their servicer, the costs to serve increases materially because those customers will gravitate to more costly service channels and they are at higher risk to take their complaints beyond the company.” From this perspective, servicers need to invest in improving the quality of their communications as they are the primary mechanism for engagement with borrowers.

“Servicers that recognize proactive customer outreach and effective problem resolution as opportunities to build stronger client relationships—rather than obstacles to overcome—are in a great position to differentiate themselves and set a new standard for the industry to follow.”

J.D. Power 2023 “Mortgage Servicer Satisfaction Study”

The need for clearer, easier-to-understand communications in financial services has also been recognized by regulators across industries and geographies. The Financial Conduct Authority in the UK and the Commerce Commission of New Zealand have now put regulations in place that put the onus on financial institutions to ensure customers can easily access information and understand what’s being communicated to them. While regulations this strict are not yet in place in the United States, the issue has been recognized by the Consumer Financial Protection Bureau (CFPB). They recently published a  letter that states that financial institutions like mortgage servicers aren’t doing enough to accommodate customers with limited English proficiency and recommends the use of more “plain language” in communications. It also recommends the use of “less dense, more digestible” content that is more conducive to digital channels and mobile devices. The (CFPB) and the Department of Justice (DoJ) have also warned servicers about a surge in borrower complaints related to “incorrect or confusing communications” and stated that, if the trend continues, they will investigate and fine the servicers who are to blame.

To adapt to this new world where clarity and consistency are a competitive and regulatory requirement, mortgage servicers need to ensure that their communications are:

1. Easy to understand: Servicers should write communications in plain language and aim for a grade six comprehension level.
2. Convey the right sentiment: Eliminate the use of overly negative words and phrasing that can easily overwhelm borrowers to such an extent that they misunderstand or ignore communications entirely.
3. Support digital channels: Simply repurposing printed communications as PDFs do not provide a good digital borrower experience, especially on mobile devices. Servicers need to optimize the style and length of their content to be appropriate for the channel being presented on.

Unfortunately, most servicers, and in turn their print service providers, rely on legacy technologies for managing borrower communications that make rectifying these issues time-consuming, costly, and a highly manual endeavor. Content authors would have to painstakingly comb through hundreds of documents and templates, one by one, to identify and correct offending content. And even if a servicer manages to bring all their content up to code, it’s likely that issues re-emerge without tools in place to maintain content standards over time.

Implementing generative AI for content optimization

There’s no question that text-based generative AI platforms like ChatGPT have the capacity to help servicers improve their content. These technologies can analyze content, rewrite it to specific reading levels and sentiment standards, as well as shorten content so that it can be repurposed on other channels. However, leveraging AI successfully depends on managing risk and having the controls in place to both ensure safety and efficiency.

Success starts with managing expectations. Think of generative AI like a new junior employee in your organization. They are smart and capable but lack detailed knowledge of your industry and your own business’s standards and practices. Like this employee, generative AI tools will only be effective in completing certain tasks and will require considerable guidance and structure before they can provide value. Making overly ambitious requests, such as writing a new letter to comply with a specific state regulation is likely unreasonable. Likewise using prompts (the directions given to a generative AI solution) that are too broad or lacking in context might lead to undesirable results. For example, you could ask ChatGPT to make a letter easier to read and convey a more positive sentiment and end up changing the underlying meaning of the letter. Avoiding these pitfalls requires using carefully tuned prompts that clearly define the task at hand, specify the output format and mitigate against known biases of the system. It’s no wonder that “prompt engineering” has emerged as an in-demand skill set as many businesses recognize its importance to their success with generative AI.

Most mortgage servicers don’t have in-house generative AI expertise yet, so it’s prudent that they explore alternative ways to harness this technology. Fortunately, there are options available beyond using these platforms directly. Some generative AI platforms have APIs available that enable third-party software vendors to build integrations with them. When leveraged within the framework and control of another application, generative AI can be an effective intelligent assistant for specific use cases which are both valuable to the user and safe for the enterprise. Here are a few advantages to consider:

1. Enterprise-grade controls: CCM systems are built to the security standards required by organizations operating in highly regulated industries. Using generative AI embedded within CCM ensures that your employees aren’t mistakenly sending customer data or sensitive content out into the public arena. You should ensure that only the content that needs to be optimized is sent and that there are controls over what is sent and when. In addition, you don’t want your content to be used to train public AI models, so ensure your CCM solution has guards in place to protect against the use of your content.
2. Higher quality output: CCM vendors with AI can standardize high-quality prompts to yield the best output consistently and save your team from having to craft them on their own.
3. Formatting and variable content preservation: Generative AI platforms are not built to handle the complex formatting and variable data content fields that are common in borrower communications. Using ChatGPT directly would mean reimplementing these requirements when bringing new content into the system every time – a time-consuming process. Look for a solution that recognizes and preserves formatting and variable data during the content optimization process which saves you from having to manually add them back.
4. A unified solution for authoring and optimization: Having to manually copy and paste content between your CCM system and the generative AI platform of choice is highly inefficient. Having a unified solution for authoring and optimization is far more convenient and efficient.

Don’t miss this chance to make a meaningful change

Generative AI presents a unique opportunity for mortgage servicers to efficiently improve the quality and consistency of their borrower communications across all channels. The safest and most effective way to take advantage is through an integration with your borrower communications management system. There’s no doubt that the generative AI landscape will continue to evolve at a rapid pace, you can rest assured that your vendor will identify innovative and applicable use cases, enabling your organization to maintain a competitive edge.