Ditch Your Income Calculation Spreadsheets: Mortgage Quality Control Just Got Smarter

Yes, this is a mortgage automation blog, but I’m not here to just brag about automation the whole way through. This is a little eye-opener—a lesson from last year and some practical guidance on what you can do if you’re planning to automate your mortgage quality control processes for faster loan approvals and higher efficiency—all with the same team you have today and the same tight budget.

So, here’s a quick sneak peek from the ACES Mortgage Quality Control Report 2024. 

In Q2 2024, the critical defect rate in mortgage loans jumped by 14.56%, reaching 1.81%—the second quarter in a row with rising defects.

Even worse, income and employment-related defects made up 37.01% of all critical defects, showing ongoing struggles with accurate borrower income verification. 

“This quarter’s rise in critical defects signals that lenders need to double down on quality control efforts, especially as volumes grow,” warned Nick Volpe, EVP of ACES Quality Management.

Did the situation get any better in Q3? We’ll get to it at the end of the blog, but do you see the irony?

2024 was celebrated as the year of AI practicality, with more companies adopting automation. Yet, the mortgage industry struggled big time. If you’re in mortgage, you know last year’s numbers weren’t great.

So, here’s the big question: Why is mortgage quality control still behind in automation, even after all these years? The first step to fixing this is knowing what’s wrong. In this blog, we’ll break down how to find the right mortgage QC tool for your business that can actually deliver results, not just fancy promises. We’ll dive into one of the biggest pain points—borrower income verification. Stick around till the end and I’ll introduce a game-changing tool that automates more than you ever imagined!

What Makes Mortgage Quality Control Challenging? 

Mortgage quality control has long been a complex and time-consuming process. With the sheer volume of documents involved, manual verification is not only tedious but also prone to errors. Over the years, automation tools like Optical Character Recognition (OCR) have attempted to streamline the process, yet they continue to fall short in one critical area—keeping up with evolving regulations and the ever-changing formats of mortgage documents. So, how can the industry move beyond these limitations? Let’s explore. 

Last year, we conducted a LinkedIn poll among mortgage QC auditors, asking what consumes most of their time in the mortgage quality control process. Over 70% of respondents pointed to cross-data verification. This comes as no surprise—traditional methods require auditors to manually compare data across multiple documents. Even automation solutions like RPA and OCR still require manual checks across different databases. 

And this is what’s been happening in mortgages. People invest in automation tools like the traditional OCR without knowing that these tools are secretly complicating the whole process with multiple checks and reviews to make sure that the data extracted is right. That’s why it's important to just automate, but automate with the right tools. 

Over 70% of the auditors say cross-data verifications take up most of their time. 

The Burden of Manual Income Calculation in Mortgage Quality Control

For mortgage auditors, income verification is a crucial safeguard. It ensures borrowers can actually afford their loans, helping to reduce the risk of foreclosures.

The 2008 mortgage crisis was a harsh reminder of what happens when income verification is overlooked. Lenders issued risky loans—many without verifying income (aka "liar loans")—which led to mass defaults and a financial meltdown.

In response, the Dodd-Frank Act introduced stricter lending standards, including the Ability-to-Repay (ATR) Rules and Qualified Mortgage (QM) Standards. These regulations made compliance a non-negotiable priority, ensuring that lenders thoroughly assess a borrower’s ability to repay before approving a loan.   

Yet, income calculation remains highly complex even today. 

Here’s why: 

1. Multiple Income Sources

• Salaried, self-employed, gig workers, and commission-based jobs—all require different verification methods.
• Some income sources (e.g., base salary) are stable, while others (e.g., commissions, freelance income) fluctuate.

2. Document Diversity

• Pay stubs, W-2s, 1099s, K-1s, tax returns, and bank statements—each follows a different format, making manual processing slow and error-prone.
• International applicants further complicate the process with currency fluctuations.

3. Debt-to-Income (DTI) Ratio Challenges

• DTI calculations require factoring in existing debt, alimony, child support, and other deductions.
• The debt-to-income ratio calculator must align with guidelines from Fannie Mae, Freddie Mac, the CFPB, and other regulators.

4. Fraud Detection

• Altered pay stubs, falsified tax documents, and inflated income statements pose significant risks.
• Detecting inconsistencies manually is inefficient and prone to human error.

Is Your Team Still Using Spreadsheets for Income Calculation?

Many mortgage professionals still rely on Excel spreadsheets for income verification. However, this approach has significant limitations:

1. High Error Rates

• A study in the Journal of Organizational and End User Computing found that 94% of spreadsheets contain errors, with an average 5.2% cell error rate.
• Miscalculations in automated mortgage processing can lead to compliance violations and loan losses.

2. Lack of Rule Configurability

• Mortgage income rules vary by borrower profile (e.g., full-time salaried vs. commission-based vs. self-employed).
• Spreadsheets lack dynamic rule configuration capabilities, making adjustments cumbersome.

3. Compliance Risks

• Automated income calculation in spreadsheets fails to keep up with regulatory changes from Fannie Mae, Freddie Mac, FHA, VA, and QM/ATR standards.
• No built-in alerts for exceeding DTI limits or missing compliance benchmarks.

4. No Real-Time Collaboration

• Multiple users working on the same file risk data overwrites or conflicting calculations.
• Spreadsheets don’t integrate with automated mortgage processing tools, requiring manual data transfer.

Why is it Crucial to Automate Borrower Income Calculation Verification? 

No matter how complex these processes are, at the end of the day, loan officers need quick, reliable numbers to determine how much a borrower qualifies for. Underwriters have to make sure those numbers are accurate and meet all risk assessment guidelines. Lenders and QC auditors are under constant pressure to ensure compliance with investor and regulatory requirements. And then there are third-party QC firms whose entire job is to double-check the accuracy of mortgage files. That’s a lot of manual work, and let’s be honest—it’s easy for mistakes to slip through. 

Infrrd Mortgage Quality Control: Verify Borrower Income Even Without Opening a Doc 

For over a decade, we’ve been working with some of the top mortgage lenders, and if there’s one thing we’ve learned, it’s this—they’re always looking for smarter ways to simplify processes without sacrificing quality. And honestly, that’s exactly how we think too.

One of our clients put it perfectly: "Our data entry teams still have to open documents and manually check if the extracted data is correct. We don’t want that. Can you build a solution where they don’t even have to open the document at all?"

One of our clients put it perfectly: Can you build a solution where they don’t even have to open the document at all?"

So, we took things up a notch. Over the years, we’ve perfected borrower income verification automation—so seamless that our clients don’t even have to touch the document. We built an entire agentic platform that doesn’t just extract data, it automates the entire process end-to-end, with little to no human intervention.

Instead of wasting hours manually verifying numbers, our AI-driven platform does the heavy lifting for you. It automates income verification, flags discrepancies instantly, and ensures every number is accurate—so you can focus on solving real problems instead of playing spreadsheet detective. 

So if you’re thinking about automating Mortgage Quality Control, just know—no one does it like we do. Here’s a step-by-step breakdown of how Infrrd’s automated borrower income calculation works.

Step 1: Document Review and Classification

First, upload your mortgage document bundle. Normally, a human reviewer would have to go through each document manually—sometimes over 100 pages—to check what’s included. But not with Infrrd’s income calculation platform. The system automatically scans the entire document package and classifies each file. It identifies whether a document is a pay stub, a VOE (Verification of Employment), a loan application (1003), or a closing disclosure. You don’t need to manually categorize anything—the system does it for you.

Once classification is complete, it also determines how many borrowers are on the loan. Instead of you having to search through the pages to figure out if there’s just one borrower or multiple, the system immediately tells you, for example, “This loan package includes one borrower named Moses and another named Maria.”

Step 2: Document Versioning

If a loan package includes multiple versions of the same document—such as different versions of the 1003 loan application or closing disclosure—the system organizes them in order. It automatically detects which version is the initial one, which one is the final, and arranges them accordingly. If you want to manually review or correct the ordering, you can do that as well.

For example, if there are three different closing disclosures, the system checks the dates and determines the latest version, the prior version, and the one before that. This helps you avoid the confusion of comparing different versions manually.

Step 3: Discrepancy Identification and Validation

Once the documents are sorted, the platform begins validating borrower details across different files. Let’s say a borrower’s first name appears as “Moses” in one document but as “Mo” in another. The system immediately flags this inconsistency as a discrepancy.

At this point, you have a few options. You can choose to ignore the discrepancy if it’s not significant, submit an issue if a correction is needed, or request a new document with the corrected name. The system allows you to categorize discrepancies based on their severity. If a critical error is found, such as mismatched borrower names on legal documents, you can escalate it directly within the platform.

Step 4: Automated Income Calculation

After verifying borrower details, the system moves on to income calculation. It extracts income-related data from pay stubs, W-2s, tax returns, and VOEs, and then performs calculations based on the type of income the borrower receives.

First, it pulls out key details like the employer name and the pay period start and end dates from the pay stubs. Even if a pay stub does not explicitly mention a start date, the system intelligently determines it by analyzing multi-version documents.

Next, it identifies the borrower’s payment frequency. If the borrower is paid biweekly, the system recognizes this and factors it into the income calculations. It also categorizes different types of earnings—such as base salary, overtime pay, and premium pay—so you can see exactly how each component is contributing to the total income.

Step 5: Trend Analysis & Discrepancy Checks

The platform doesn’t just calculate income; it also tracks trends over time. If a borrower’s income fluctuates significantly from year to year, it flags it as a potential issue.

For example, if a borrower’s income increased by 27% in 2022 and then by another 16.5% in 2023, the system detects that growth pattern. If the increase or decrease is above a predefined threshold—such as 10%—it highlights it as a discrepancy for further review.

Let’s say a borrower’s overtime pay declined and then jumped by 30%. The system flags this variation as something that may need additional verification. You can configure these discrepancy rules based on your requirements.

Step 6: Post-Closing Audit

This tool is also valuable for post-closing audits. Since the loan is already closed, the system helps identify any potential issues that were missed during underwriting.

For instance, it checks whether the borrower’s pay stub was within an acceptable timeframe at the time of closing. Just like how a visa application requires recent bank statements, mortgage lenders also require pay stubs to be within a certain period from the loan closing date. The system automatically validates these rules, ensuring compliance with investor and regulatory requirements.

Step 7: Dynamic Review & User Adjustments

Once all the calculations and verifications are complete, the platform allows you to dynamically review the results. Think of it as a live spreadsheet. If you disagree with a number or want to explore different scenarios, you can manually adjust the figures, and the system will instantly update the calculations.

For example, if a borrower’s biweekly salary was calculated at $2,500, but you believe an adjustment is needed, you can change the value and see the impact on monthly and annual income in real-time. This flexibility allows you to simulate different income scenarios and make informed decisions before finalizing the numbers.

Step 8: Integration & API Support

The platform is designed to work seamlessly with Loan Origination Systems (LOS) and other existing mortgage software. You can choose to use it as a standalone interface, or integrate it into your existing underwriting workflow through APIs.

For example, if a lender prefers to keep their current document review system but wants to enhance income verification, they can integrate Infrrd’s API to automatically calculate income and detect discrepancies without switching platforms.

Step 9: AI-Powered Assistance (Future Enhancements)

Soon, the platform will include an AI assistant that allows you to interact with the system conversationally. Instead of manually navigating through dashboards, you’ll be able to ask the assistant questions like:

• “What discrepancies were found in this file?”
• “Can you summarize the borrower's income history?”
• “Show me any missing documents for this loan.”

The assistant will also provide intelligent recommendations to help users prioritize their work. If there are multiple discrepancies, it will suggest the most critical ones to address first.

Coming back to the ACES Mortgage Quality Control Trends reports, the Q3 results brings a mix of progress and persistence. The critical defect rate dropped 17% from Q2 to Q3, settling at 1.51%—a clear indication of improvement. But while the numbers improve, they also highlight the ongoing challenges we need to tackle.

Income/Employment defects remain the top issue at 25%, followed by Assets (16.67%) and Credit & Loan Documentation (12.12% each). These persistent problem areas serve as a reminder that while we’re making headway, achieving airtight mortgage quality control is still a work in progress.

This isn’t just about compliance—it’s about trust, accuracy, and strengthening the foundation of the mortgage industry. These insights should serve as both a milestone and a call to action to refine processes, tighten controls, and build a more resilient Mortgage QC landscape in 2025.

The progress is real—but so is the work ahead. 

‍