Ending the hire/layoff cycle in mortgage lending

The mortgage origination industry has long made use of an operational strategy that relies on “staffing up” during cycles of high volume and then making staff reductions, sometimes en masse, when origination volume declines. While this was once a necessary business strategy, the technology now exists to implement a much more efficient approach. In order to achieve this, lenders need to take a thoughtful approach to the implementation of technology, and the very highest decision makers need to push their orgs to adapt processes to fully realize technology’s benefits.

Some credit is due to the mortgage industry for having made real progress in a few short years with regard to making automation a priority. Structured data sources, automated underwriting tools, and task and workflow-based systems can automate and accelerate many of the functions traditionally performed by human employees. The technology to do this is possible and in many cases, has even been built – it just hasn’t been properly implemented. We have the means today to shift lenders to far more efficient staffing and personnel models. 

The problem lies in a simple insight: cost and labor doesn’t come out of the process when you automate a step, but rather when the person stops doing that step.

While we have adopted technologies and “turned them on” across the industry, so to speak, we have not re-architected our processes around them. In order to do that, I suggest 3 truths that lenders should keep in mind:

1. There are some things that technology can simply do better than humans – that work needs to be moved to technology.
2. There are other things that humans still need to do, but the more that companies can make those tasks “lower skill” and easier to train around, the more scalable and nimble the organization can be as the market changes.
3. It is very easy to add steps to a process, and overwhelmingly difficult to subtract them.

Moving workloads from people to technology

As an industry, we need to start by establishing a basic acceptance that technology simply does many things better than humans can. For example, technology is more reliable at making sure a flood report is ordered as soon as a loan is moved to processing than a human with a checklist is – it’s a simple rule that fires every time. This is an easy thing to agree on when technology does things “perfectly”, but what about when the technology has an error rate? For example, does OCR have to be 99%, 99.9%, or 100% right in extracting the first name off a driver’s license before we decide we can have people stop doing it? While a simple solution might be to compare the computer’s error rate to the human’s, and switch when computers make fewer mistakes, we all know this is difficult in practice – one only needs to look at the regulatory backdrop around self-driving cars as a prime example in another industry. Regardless of whether a company is willing to take a more cutting edge approach (switching as soon as the technology is “good enough”) or would rather be a later adopter, it is critical that lenders have a strong framework and strategy around when they shift work from humans to technology.

If we assume for a moment that we are willing to move work to technology once it can do it better than humans, the next step is to survey the market and understand what “automation” capabilities exist, and to what extent they can replace the entire task a human is doing, not just a pretty demo. This requires not just a deep understanding of the technology, but also an understanding of the current human process – is the same person ordering flood reports and reviewing the results? Is that person also troubleshooting when the flood vendor is unable to return a successful result because the property address hasn’t been verified yet? For many mortgage lenders, either very robust automation capabilities are necessary to replace these large, complex tasks their operations people are really doing, or their process needs to be simplified and broken down significantly into smaller “jobs to be done” so that simpler automation solutions can take those jobs on.

Making the work people still have to do more efficient

Once we’ve looked at work that can be moved to technology, the state of the industry today suggests that there surely will still be work for people as well; how can we make that as efficient as possible? There are two vectors lenders should think about when they think about the efficiency of their people: how much “work” (measured in loans, or ideally in tasks) one person can get done in a unit of time when fully ramped, and how long it takes to ramp them.

Work per user per unit of time is measured all over our industry: how many files can an underwriter review a day? How many loans can a processor manage in their pipeline at once? When thinking about improving the efficiency of work that people are doing and that can’t be moved to technology, lenders often focus on allowing those people to focus (there are elaborate solutions, like workflow systems, and simple solutions, like not letting your loan officers instant message your underwriters). Continuously re-evaluating how your technology is serving or hindering your team’s ability to work efficiently (system latency, easy navigation) is critical.

Less obvious, though – for work that still needs to be done by people, it’s also worth asking how we can make that work more intuitive, easier to train on, and split in a way that the bulk of it can be done by lower skilled workers. Making the work “simpler” or “easier” allows for improved elasticity in the business; rather than hiring specialists, lenders can move more general workers around to cover bottlenecks in their manufacturing process, or even slide employees between origination and servicing if the jobs and systems are really made simple enough. When a mortgage business employs a strategy that decreases the need for skill, the next market boom will require that business to simply flex 10 or maybe 20 people from other areas of the business into production to assist with the sustained volume. Instead of having to hire (and later, lay off) 100 new people and train them, the business can accommodate demand without artificially bloating its workforce.

Actually changing the process with subtraction, not just addition

One of the hardest things to do in any organization is to subtract things – adding a weekly meeting is almost frictionless in most companies, but removing it requires a brave individual to speak up, and then near-total consensus from the attendees. Similarly, in loan manufacturing, lenders add steps (things to check, documents to collect or send out) often, but almost never inventory all of the work their teams are doing and figure out what doesn’t need to be done anymore.

Compounding this, “redundant work” is actually commonly employed as a strategy in mortgage origination to ensure that files are high quality – if 2 people each have 10% error rates on missing a document in the file, for example, making sure that one checks it, and then the other checks the first person’s work, will allow for a 1% error rate. This simple fact often leads to checkers checking checkers.

This means that people in the mortgage origination process often do work on a file, expecting that work to have been done already anyways, and therefore that automating that work upfront doesn’t fundamentally change their behavior. In order to see the full benefits of any kind of work shift to automation and technology, the entire team downstream of that work needs to be retrained no longer to check that step. So how can this be done?

First and foremost, executive buy-in is critical to see any reduction of steps; very rarely do individual contributors feel empowered to remove steps from the process. Tobi Lutke, CEO at Shopify, has a famous quote that “The best thing founders can do is subtraction. It’s much much easier to add things than it is to remove.” Only executives and leaders have the social capital necessary to actually get people to stop doing stuff.

Once an executive has decided to cut a bunch of steps from the process, and ideally continue to revisit and cut steps as a continuous, living process, there are two ways to actually implement this. One is to have an extremely strong operations group that continuously trains and retrains; measuring the effectiveness of the team and ensuring an extremely high degree of compliance with written and documented standard operating procedures. 

The second option to implement effective change-management on your automation journey is to encode the standard operating procedure inside of the system being used to originate the loans so that the software manages the process. When that happens, change management becomes something akin to a software update (with which people will comply because they are only doing the work the software directs them to do) instead of a complex and cumbersome re-training. The change management is, essentially, “baked into” the infrastructure itself.

Regardless of whether an operational or technical solution to this problem is employed, all of the automation in the industry won’t save money if each lender doesn’t individually develop a strategy to achieve the all-important goal of subtraction.

Tying these things together

We’ve long known that the mortgage industry struggles with a lack of elasticity and scalability. The market conditions and unique events of the past four or five years have starkly highlighted that problem. As our market has lurched from boom to bust in response to pandemic and high interest rates, we’ve seen an almost continuous parade of hiring and reductions-in-force. That, in turn, has had a broad impact on work culture across the industry, including a reduced level of loyalty industry-wide and a volatility that makes it nearly impossible to plan much further than the next peak or trough in the market cycle. 

To actually get to a world where scaling production up and down doesn’t require swings in headcount, the marginal cost of labor inside each loan needs to be as low as possible, and then any need to “staff up” or “staff down” needs to be low training enough to allow the repurposing of existing staff instead of hiring and firing of staff outside the company. Getting to this world requires a careful re-litigation of existing processes, with executive buy-in, to subtract unnecessary steps and simplify the process, and this is an exercise that must be done continuously by lenders. This is not simple – there is no silver bullet solution – and it is hard work, but it is one of the greatest problems mortgage originators face today, and well worth the work.

As CEO at Vesta, Mike leads sales, product development, and implementations as the team redefines origination platforms for modern lenders. Previously, Mike spent 4 years on the early product team at Blend, where he launched key components of the flagship mortgage platform, and later started and ran new business lines such as Blend Insurance. Mike graduated from Stanford University with an MS in Computer Science/AI and a BA in Economics.

This column does not necessarily reflect the opinion of HousingWire’s editorial department and its owners.

To contact the editor responsible for this piece: [email protected]