You submit your mortgage application online at 9 a.m. By 9:04 a.m., you have an answer. No phone call, no waiting three business days for a loan officer to review your file manually. Just a clean digital decision: approved, pending conditions.
Most borrowers experience this and move on. What almost nobody stops to ask is: what just happened? Who made that call, what data did they use, and — critically — does that decision have anything to do with the interest rate you were just quoted?
The answer to that last question is no. And understanding why that answer is no could save you tens of thousands of dollars over the life of your loan.
Automated underwriting systems, or AUS, are the engines that power nearly every mortgage decision made in the United States today. They are fast, consistent, and largely invisible to borrowers. But the AUS approval is only half the story. The rate you pay is determined by a completely separate pricing mechanism — one where the lender you choose, and how many wholesale investors they can access, matters enormously.
This article breaks down exactly how automated mortgage underwriting systems work, what they evaluate, what they deliberately do not touch, and why the same borrower with the same AUS outcome can receive meaningfully different rates depending on which lender’s shelf their application lands on.
By Duane Buziak, NMLS #1110647 | Coast2Coast Mortgage LLC NMLS #376205
The Engine Behind the Decision: What AUS Actually Does
When your application enters the automated underwriting system, it is being evaluated by one of a small number of dominant engines — each tied to a specific loan type and investor.
For conventional loans sold to the secondary market, the two engines are Fannie Mae’s Desktop Underwriter (DU) and Freddie Mac’s Loan Product Advisor (LPA). Government-backed loans run through their own agency-specific systems: FHA loans use the FHA TOTAL Scorecard, and VA loans are evaluated through VA-specific underwriting guidelines referenced in the VA Home Loans program. USDA loans have their own automated system as well.
Each engine ingests a standardized set of data points from your loan application and credit report. The core inputs include your credit profile, debt-to-income ratio (DTI), loan-to-value ratio (LTV), asset reserves, employment history, and property type. The system weighs these factors against each other — not as a simple checklist, but as an interconnected risk model. A borrower with a lower credit score might still receive an approval if their DTI is strong and they carry substantial reserves.
One detail worth understanding in 2026: Fannie Mae and Freddie Mac have been transitioning away from classic FICO models toward newer scoring models, including VantageScore 4.0 and FICO 10T, as announced by the Federal Housing Finance Agency. This transition affects which version of your credit score the AUS actually evaluates — and it is worth confirming with your broker which model applies to your specific loan type at the time of application.
The output of the AUS is a risk classification, not a rate. DU produces findings such as Approve/Eligible, Approve/Ineligible, Refer/Eligible, or Refer with Caution. LPA uses Accept, Caution, or Ineligible. These findings tell the lender whether the loan meets agency guidelines and what documentation conditions must be satisfied to close.
Here is the distinction that most borrowers never learn: the AUS does not set your interest rate. It determines eligibility and conditions. Rate pricing happens through an entirely separate mechanism called the Loan-Level Price Adjustment grid — and that grid is where the real money is made or lost.
Think of AUS as the bouncer at the door. It decides who gets in. What you pay once you’re inside is a completely different conversation.
LLPAs: The Pricing Layer the AUS Doesn’t Touch
Once your application clears the AUS with an Approve/Eligible or Accept finding, the lender turns to a pricing grid to determine your actual interest rate. For conventional loans, this grid is Fannie Mae’s Loan-Level Price Adjustment matrix — a publicly available table of fee adjustments that stack on top of the base rate based on specific characteristics of your loan.
The primary axes of the LLPA grid are: credit score band, loan-to-value ratio, loan purpose (purchase versus cash-out refinance), occupancy type (primary residence, second home, investment property), property type (single-family, condo, multi-unit), and product type. Each combination of these variables produces a specific price adjustment, expressed in points — where one point equals one percent of the loan amount.
Here is where it becomes concrete. Two borrowers both receive an Approve/Eligible finding from Desktop Underwriter. Both are purchasing a primary residence with a conventional 30-year fixed at 80% LTV. The only difference is their credit score: Borrower A has a 680 FICO, and Borrower B has a 740 FICO. According to Fannie Mae’s published LLPA matrix, these two borrowers face meaningfully different price-adjustment hits. Borrower A’s LLPA stack is higher, which translates directly into a higher note rate or additional closing costs — even though both borrowers passed the same AUS.
This is the core mechanic most borrowers miss entirely. The AUS approval is binary: you either pass or you don’t. The rate is continuous: it shifts with every variable in your LLPA profile. Two borrowers can have identical AUS outcomes and face rates that differ by 0.375% or more, purely because of where they land on the LLPA grid.
Now here is where the broker structure becomes structurally significant — not as a marketing claim, but as a mechanical advantage. A wholesale mortgage broker submits the same AUS file to a wide shelf of wholesale investors and can identify which investor’s LLPA pricing is most favorable for that specific borrower profile. A single-shelf retail lender — including large retail operations like Rocket or Movement — prices against their own internal grid only. They have no mechanism to shop your LLPA stack across competing investors.
This is not a personality difference. It is a structural one. The broker’s value is LLPA arbitrage across a wholesale shelf. The retail lender’s constraint is that their shelf has exactly one option: their own.
For borrowers with profiles that carry elevated LLPAs — moderate credit scores, higher LTVs, cash-out refinances, condo purchases — this difference is not marginal. It is the difference between the best available pricing for your file and whatever a single lender happens to offer that day.
The Math That Makes It Real: A Worked Rate-Comparison Example
The following is a hypothetical illustration using realistic market parameters. It is intended to demonstrate the mechanical relationship between LLPA pricing and borrower cost — not to guarantee any specific rate or outcome.
Let’s build a specific borrower profile: $400,000 loan amount, 90% LTV (purchase price approximately $444,444), 700 FICO, primary residence, conventional 30-year fixed. This borrower has cleared AUS with an Approve/Eligible finding from Desktop Underwriter. The AUS outcome is identical regardless of which lender receives the file.
At 90% LTV with a 700 FICO on a conventional loan, this borrower carries a meaningful LLPA stack. The 90% LTV and the 700 credit score band each contribute price adjustments that push the effective rate higher than what a 760 FICO borrower at 80% LTV would see. The AUS does not care about this distinction. The pricing grid does.
Scenario A — Single Retail Lender: The lender prices this file at 7.25% on a 30-year fixed. Monthly principal and interest payment on $400,000 at 7.25%: using the standard amortization formula P&I = [P × r(1+r)^n] / [(1+r)^n – 1], where P = $400,000, r = 0.0725/12 = 0.006042, and n = 360, the monthly payment is approximately $2,729.
Scenario B — Wholesale Broker Access: After shopping the file across multiple wholesale investors, the broker identifies an investor whose LLPA grid prices this specific profile at 7.00%. Monthly P&I on $400,000 at 7.00%: using the same formula with r = 0.07/12 = 0.005833, the monthly payment is approximately $2,661.
Monthly savings: $2,729 minus $2,661 = $68 per month.
Now apply the breakeven calculation. Suppose the lower rate in Scenario B carries a 0.5-point origination cost — $2,000 on a $400,000 loan. To determine when the lower rate pays for itself, divide the additional closing cost by the monthly savings: $2,000 divided by $68 equals approximately 29 months. If you stay in the home or keep the loan beyond 29 months, Scenario B is the better economic choice in total dollars paid.
Over a full 30-year term, the difference compounds further. At $68 per month across 360 payments, the total savings in Scenario B exceeds $24,000 — before factoring in the accelerated principal paydown that comes with a lower rate.
The critical point: the AUS outcome was Approve/Eligible in both scenarios. The borrower’s credit profile, DTI, and assets were identical. The only variable was which investor’s LLPA grid the file was priced against. This is the rate-shopping argument expressed in mechanical terms, not marketing language.
This is also why a no-credit-impact mortgage pre-approval — run before any hard inquiry is triggered — is the correct first step. You want to understand your AUS profile and your LLPA stack before you commit to any single lender’s pricing.
When the Algorithm Says ‘Refer’: Manual Underwriting and Edge Cases
A Refer or Refer with Caution finding from the AUS is not a denial. It is the system’s way of saying: I cannot approve this automatically, but I am not saying no either. A human underwriter needs to evaluate this file.
Manual underwriting is the process by which a trained underwriter reviews compensating factors that the AUS model either cannot capture or weighs too conservatively. Strong compensating factors typically include a low payment shock from current housing costs, substantial liquid reserves, a long history of on-time rent payments, or a stable employment record in the same field for many years.
The borrower profiles most likely to encounter a Refer finding include: thin credit files with limited trade line history, self-employed borrowers or 1099 workers whose income documentation is complex, borrowers with recent credit events such as a late payment or short sale, and borrowers whose debt-to-income ratio approaches or exceeds standard AUS thresholds. None of these profiles are automatically disqualifying — they simply require a human review that the automated system is not designed to provide.
VA loans are particularly notable here. Per the VA Lenders Handbook, Chapter 4, VA guidelines allow for manual underwriting with compensating factors for borrowers who receive a Refer finding — making VA loans one of the most flexible products available for borrowers with non-traditional credit profiles. The VA does not impose a hard minimum FICO requirement in the same way conventional guidelines do; rather, the focus is on the full picture of the borrower’s creditworthiness.
Before any AUS run — automated or manual — the correct first move is a soft-pull pre-qualification. A no-credit-impact mortgage pre-approval, like the NoTouch Credit Pull offered through Shop Mortgage Rates, allows a broker to review your credit profile, model likely AUS outcomes, and position your file optimally before any hard inquiry appears on your report. This matters because multiple hard inquiries in a short window, while treated as a single inquiry for mortgage rate-shopping purposes by most scoring models, can still create friction in the process. Starting with a soft pull gives you a clear picture of where you stand — and which AUS engine and investor shelf is most likely to produce the best outcome for your specific profile.
Broker vs. Single-Shelf Lender: AUS Access and Rate Outcomes
The structural differences between a wholesale mortgage broker and a single-shelf retail lender are not a matter of opinion. They are a matter of what each entity can and cannot do with your loan file. The table below makes this concrete.
AUS Engines Available
Wholesale Broker (Shop Mortgage Rates): DU, LPA, FHA TOTAL Scorecard, VA AUS — full access to all major agency engines across multiple investors.
Single-Shelf Retail Lender (e.g., Rocket, Movement): Typically one proprietary system plus one agency engine, constrained to their own product shelf.
National Rate Aggregator: No actual lending capability — these are lead-generation platforms that refer your information to lenders. They do not run AUS or originate loans.
Wholesale Investor Access
Wholesale Broker: 500+ wholesale investors, each with their own LLPA pricing grid.
Single-Shelf Retail Lender: Own product shelf only — one pricing grid, one set of overlays.
National Rate Aggregator: Not applicable — no origination capability.
LLPA Pricing Flexibility
Wholesale Broker: Can identify and select the investor whose LLPA grid produces the most favorable pricing for the borrower’s specific profile.
Single-Shelf Retail Lender: Fixed to house pricing — no mechanism to shop LLPA grids across competing investors.
National Rate Aggregator: Not applicable.
Investor Overlay Flexibility
Wholesale Broker: If one investor’s overlay conflicts with the borrower’s profile (e.g., a minimum credit score requirement stricter than the AUS guideline), the broker can move the file to an investor without that overlay.
Single-Shelf Retail Lender: Bound by their own overlays — an Approve/Eligible AUS finding does not guarantee the lender will fund the loan if their internal overlay is more restrictive.
National Rate Aggregator: Not applicable.
Manual Underwriting Option
Wholesale Broker: Available through multiple investors, increasing the likelihood of finding a path to approval for Refer findings.
Single-Shelf Retail Lender: Depends entirely on that single lender’s policy — some retail lenders have limited manual underwriting capacity.
National Rate Aggregator: Not applicable.
Wholesale Broker (Shop Mortgage Rates): Yes — NoTouch Credit Pull, a mortgage pre-approval without a hard pull, available before any AUS submission.
Single-Shelf Retail Lender: Varies; many require a hard credit pull before providing any meaningful rate information.
National Rate Aggregator: Typically a hard pull or lead-generation capture — your data is sold to multiple lenders, not used for actual pre-qualification.
The investor overlay point deserves particular emphasis. Many lenders impose credit score minimums, DTI caps, or reserve requirements that are stricter than what the AUS guidelines require. An Approve/Eligible finding from Desktop Underwriter means Fannie Mae’s model approved the loan — it does not mean every lender will fund it. A broker can identify which investors on their shelf have the most favorable overlays for a given borrower profile and route the file accordingly. A single-shelf lender has no equivalent option.
Eight Questions Borrowers Actually Ask About Automated Underwriting
Does running an AUS check hurt my credit score?
Running the AUS itself does not hurt your credit score. The AUS uses credit data that has already been pulled. What affects your score is the hard credit inquiry that occurs when a lender formally pulls your credit report. Starting with a soft credit pull mortgage pre-qualification — which does not trigger a hard inquiry — allows you to understand your AUS profile before any credit impact occurs. The CFPB confirms that soft inquiries do not affect credit scores.
What credit score model does Desktop Underwriter use?
Fannie Mae’s Desktop Underwriter has historically used classic FICO score models. However, Fannie Mae and Freddie Mac have been transitioning to newer models, including VantageScore 4.0 and FICO 10T, following FHFA’s approval of these models. As of mid-2026, you should confirm with your broker which scoring model applies to your specific loan type, as the implementation timeline has been phased.
What does a ‘Refer’ finding mean on my mortgage application?
A Refer finding means the AUS could not automatically approve the loan, but it is not a denial. It signals that a human underwriter needs to review the file and evaluate compensating factors. Many borrowers who receive a Refer finding are ultimately approved through manual underwriting — especially for VA and FHA loan types, which have robust manual underwriting pathways.
Can I still get a mortgage if my AUS comes back as Refer?
Yes, in many cases. A Refer finding opens the path to manual underwriting, where a human evaluates your full financial picture rather than relying solely on the automated model. Borrowers with strong compensating factors — substantial reserves, low payment shock, stable long-term employment — are frequently approved through manual underwriting even after a Refer finding. VA loans in particular allow manual underwriting with compensating factors per the VA Lenders Handbook.
What is the difference between DU and Loan Product Advisor?
Desktop Underwriter (DU) is Fannie Mae’s AUS, while Loan Product Advisor (LPA) is Freddie Mac’s equivalent. Both evaluate conventional loans against their respective agency guidelines, but they use different risk models and may produce different findings for the same borrower. A broker can run the same file through both systems and select the more favorable outcome — a single-shelf lender typically cannot.
Can I see my own AUS findings report?
You have the right to request your AUS findings from your lender or broker. The findings report is a formal document generated by the AUS that lists the approval decision, the conditions required for closing, and the data points the system evaluated. Ask your loan officer for a copy of the DU findings or LPA feedback certificate — any reputable broker should provide this without hesitation.
Does the automated underwriting system determine my interest rate?
No. This is one of the most important distinctions in mortgage lending. The AUS determines eligibility and conditions — whether you qualify for the loan. Your interest rate is determined separately through the Loan-Level Price Adjustment grid, which prices your loan based on credit score, LTV, loan purpose, occupancy, and other factors. Two borrowers with identical AUS outcomes can receive materially different rates depending on their LLPA profile and which lender prices their file.
How long does an AUS approval stay valid?
AUS findings are typically valid for 120 days from the date the credit report was pulled, though this can vary by loan type and investor. If your loan does not close within that window, the file may need to be resubmitted through the AUS with a refreshed credit report. Changes to your financial profile between the original AUS run and closing — new debt, a job change, a significant credit event — can affect the outcome of a resubmission.
Putting It All Together: Your AUS Profile Is the Starting Point, Not the Finish Line
The core insight of this article is simple but consequential: the AUS decision and the rate you are offered are two separate outputs of two entirely separate systems. An Approve/Eligible finding is a floor, not a ceiling. It tells you that you qualify. It tells you nothing about whether you are getting the best available rate for your specific profile.
What determines your rate is the LLPA grid — and the LLPA grid you are priced against depends entirely on which lender receives your file. A broker with access to 500+ wholesale investors can identify the investor whose pricing grid is most favorable for your specific combination of credit score, LTV, loan purpose, and property type. A single-shelf retail lender prices against one grid and calls it a day.
The right first move is always a no-credit-impact mortgage pre-qualification. The NoTouch Credit Pull lets you understand your AUS profile, your likely LLPA stack, and your realistic rate range — before any hard inquiry touches your credit report and before you are locked into any single lender’s offer.
Your dream home is within reach. Discover how much you could save with personalized mortgage rates tailored to your unique situation. Securely pre-qualify in minutes with no impact to your credit score and compare competitive offers from trusted lenders who are ready to help you save.