How IMBs Can Cut Origination Cost Per Loan Below $11,000

TL;DR

Independent mortgage banks spent $11,230 to originate a single loan in Q4 2024, according to Freddie Mac. That number has not fallen despite years of LOS, POS, and CRM investment.

The cost sits in four places. Manual borrower follow-up, condition clearing, application fallout, and processor file work each drain hours that software was supposed to eliminate.

This guide maps four cost drivers to four automation levers and one framework for choosing tools.

Why Origination Costs Are Still Above $11,000 Per Loan

The $11,230 per-loan figure from Freddie Mac's Q4 2024 study is the industry's reference point, and it has barely moved. IMBs bought loan origination systems, point-of-sale portals, and CRMs expecting that number to drop.

It didn't. Those tools digitized the paperwork but left the manual labor intact.

A borrower still gets a phone call to chase a missing pay stub. An LO still relays condition lists by hand. A processor still sorts and labels documents before anyone can review the file.

Software moved these steps onto screens without removing the human effort behind them. The borrower-facing work that drives most of the $11,230 stayed manual.

Where Origination Cost Actually Lives

The $11,230 per loan in the Freddie Mac Q4 2024 benchmark does not sit in one line item. It accumulates across four operational steps where staff spend hours on work software was supposed to remove. Find these four drivers in your own pipeline and you find the cost.

Manual borrower follow-up. Loan officers and processors chase documents by phone and email, one outreach at a time. A single borrower can absorb a dozen calls before a file is complete. Those hours are billable staff time spent waiting on paperwork.

Condition clearing bottleneck. When AUS output returns a condition list, the LO becomes the relay between that system and the borrower. Every condition gets read, translated, explained, and re-checked by hand. The borrower waits on the LO, and the LO waits on the borrower.

Borrower fallout and abandonment. Applications that stall mid-funnel are pure sunk cost. You paid for the credit pull, the staff time, and the marketing that sourced the lead, then earned zero revenue when the borrower walked. Slow follow-up is the most common reason they leave.

Processor file assembly and routing. Before anyone can review a file, a processor sorts, labels, and routes incoming documents. Mislabeled uploads and missing pages send the file back to the borrower, restarting the clock.

Each driver shares one trait. The work is repetitive and borrower-facing, and your LOS, POS, and CRM digitized the form without removing the labor. The next section maps a specific lever to each one.

Four Levers to Cut Per-Loan Origination Cost

Each cost driver maps to a specific operational change. Four levers, each tied to one of the drivers above.

Automate Borrower Outreach Across Channels

Manual follow-up eats LO and processor hours, and it maps directly to the $11,230 per-loan cost Freddie Mac reported for Q4 2024. Scheduled outreach across SMS, voice, and email replaces the staff-initiated phone tag that drives most of that labor.

A loan officer chasing one document might make three calls and send two emails over a week. A sequenced campaign covers the same borrower automatically and escalates only when a response stalls.

That collapses a dozen manual touch points into a handful of staff interventions, and it frees your team for files that actually need a human.

Shift Document Collection From Portal-Passive to Agent-Initiated

A document portal waits. It lists outstanding items and sits there until the borrower decides to log in, find the right file, and upload it. Most borrowers don't.

An agent prompts. It tells the borrower exactly which document is missing, sends the request through SMS or email, and follows up when nothing arrives.

That difference shows up in cycle time. Portal-passive collection stretches the document phase for days while files sit untouched. Agent-initiated collection compresses it because the borrower never has to remember on their own.

Route Conditions Directly to Borrowers Without LO Intermediation

Every condition the AUS returns currently routes through the LO. The borrower asks what the request means, the LO explains, the borrower responds, and the LO relays it back. Each loop burns hours that show up in the $11,230 Q4 2024 Freddie Mac cost figure.

Automation closes that loop directly. A tool reads the raw condition output, translates it into plain language, and sends the request straight to the borrower.

• Borrower gets a clear, specific ask instead of underwriting shorthand
• The LO drops out of routine relay work
• Response tracking happens automatically, so nothing stalls unseen

Validate Documents at Upload, Not at Review

Upload-time validation checks a document the moment a borrower submits it. It confirms three things before the file moves forward.

• Legibility so an underwriter can actually read the page
• Completeness so no pages are missing from a multi-page statement
• Type match so a W-2 lands where a W-2 belongs

Catch these at upload and you avoid the re-work loop. A processor finds the problem days later, requests a corrected file, and waits for the borrower to respond. Each cycle adds days and staff hours to a per-loan cost already above $11,230.

What to Look For in AI Mortgage Automation Tools

Most AI mortgage tools demo well and integrate poorly. Judge them on four operational questions, not their feature lists.

Overlay vs. rip-and-replace. A tool that overlays your existing LOS and POS earns its keep on day one. A tool that demands a platform migration costs you months before it touches the $11,230-per-loan problem.

• Ask whether it connects to your current stack or requires you to abandon it.

Proactive vs. reactive. A reactive tool waits for the borrower to log in and act. A proactive agent initiates the outreach, the document request, and the follow-up without staff prompting.

• The reactive version automates nothing. It just relocates the manual work.

Channel coverage. Borrowers respond on different channels at different times. A tool that only sends email misses the borrower who answers a text in seconds.

• Confirm it handles SMS, voice, and email, not a single modality.

Integration depth. Surfacing a notification is not the same as moving a loan forward. A tool that reads and writes loan data can update conditions and statuses directly.

• A tool that only reads data still leaves a human to key in the response.

Score each vendor against these four points before you weigh price. A tool that fails on integration depth or proactivity will not move your cost per loan, regardless of how the interface looks.

How Penny Addresses Each Cost Driver

Penny, built by Copperlane, is one tool that maps to the four levers above. It sits on top of your existing stack and works the borrower-facing steps that drive most of the $11,230 per-loan cost in the Q4 2024 Freddie Mac figure.

Here is how it addresses each cost driver.

• Manual follow-up. Penny runs SMS, voice, and email outreach sequences autonomously. No LO has to initiate the chase or schedule the next call.

• Condition clearing. Penny reads AUS conditions and rewrites them as plain-language borrower requests. It tracks responses and follows up without pulling the LO into every exchange.

• Borrower fallout. When an application stalls, Penny sends proactive prompts to re-engage the applicant before they abandon. That recovers files that would otherwise become sunk cost with zero revenue.

• Processor file work. Penny validates documents at upload for legibility, completeness, and type match. Issues surface before the file reaches a processor, so the file arrives review-ready instead of triggering re-work.

Be clear about what Penny is not. It is not a loan origination system, a point-of-sale platform, or a decision engine. It does not replace your LOS or make credit calls.

Penny handles the borrower communication and document work that those systems leave to your staff. You keep the platforms your team already knows and add an agent that does the manual labor underneath them.

If you want to scope where Penny fits in your current workflow, Copperlane can map it to your specific cost drivers before you commit to anything broader.

Start With a Pilot, Not a Platform Swap

Do not rip out your stack to test automation. Pick one loan channel or one branch and run the levers there first.

Measure two numbers before you start. Track cost per loan and average cycle time, then compare the same two metrics after the pilot runs. That gives you a defensible number against the $11,230 Q4 2024 Freddie Mac benchmark, not a vendor promise.

Scope a pilot with the team at Copperlane. Bring your current per-loan cost and one channel you want to test, and run it for a contained window.

Frequently Asked Questions

What is the average cost to originate a mortgage loan? Freddie Mac put origination cost at $11,230 per loan in Q4 2024. That figure covers staff time, technology, and overhead across the full origination cycle. Most of it sits in manual borrower-facing labor.

What is the biggest driver of high origination costs for IMBs? Manual borrower follow-up is the largest single drain. Loan officers and processors spend hours chasing documents and relaying conditions. Automating those touch points cuts the labor that drives per-loan cost.

Can AI tools integrate with existing LOS platforms without a full replacement? Yes. Overlay tools like Penny connect to your current LOS and POS through integrations. You avoid a platform migration and keep your existing system of record.

How long does it take to see cost reduction after deploying borrower automation? Cycle time usually drops within the first loan cohort. Measure cost per loan before and after the pilot to confirm the gain.

What is the difference between a proactive AI agent and a borrower-facing portal? A portal waits for the borrower to log in and act. A proactive agent like Penny initiates outreach across SMS, voice, and email. The agent prompts borrowers instead of hoping they return.