Rates broke lower on a Tuesday. By nine your top producer is pulling the funded book out of the LOS. By eleven the export is a mess — the rate column ignores the seller-paid buydown, half the loans transferred servicing, and nobody can say which files are still inside the early-payoff window. At noon a past client texts: "Got a call from somebody saying I could save $340 a month. Is that real?"
It was real enough. Somebody worked your book before you did. They had no relationship, no file, no memory of the eleven weeks you spent getting that loan to the table. They had a list, and you didn't.
The short version
- Your funded book is the only lead source you already paid for. Work it on a schedule, not when rates move.
- Sort triggers into three lanes — what the data watches, what the calendar watches, what only a call finds. Automate the first two.
- Run the arithmetic before you dial, and show the client all of it, including the term that pays you less.
- Every candidate clears four gates: EPO window, horizon, net benefit, and whether you'd sign it yourself.
- Open as an annual review. The call where you say "don't do this" is the one that buys the next twenty years.
Somebody is already working your book. It just isn't you.
The neglect is structural, not lazy. LO comp pays on the next funding, so attention moves the day the wire clears. No job description in the branch says "look backward." Servicing usually transfers inside sixty days, so the statement that lands in the mailbox carries someone else's logo and someone else's phone number.
Meanwhile, a mortgage is public record. The day the deed records, the amount, the lender, and the date are visible to anyone who wants to buy that data, and plenty of shops do. When rates move, your funded clients get called by people who know their balance and nothing else about them.
That is the whole competitive picture. They have the list. You have the relationship, the tax returns, the story about the aunt who co-signed, and the fact that this borrower answered your calls for six straight weeks once. None of it matters if you never dial.
The trigger board: three lanes, not one list
Most funded-book programs fail because they run on one signal — rate gap — and rate gap is the only trigger a purchased mailing list also has. You will never win a race you're running on your competitor's data. Sort your triggers into three lanes instead, each with its own owner and cadence.
Lanes one and two are watch-work, and watch-work belongs to software. Lane three is the entire reason the annual call exists. A divorce, a business started, a kid heading to school two states away, a rental they've been circling for a year — none of that appears in an export, and all of it changes the answer.
One rule for lane one: set your rate-gap threshold and defend it. A quarter point is not a conversation. Three quarters, on a clean file with real costs, usually is.
Run the arithmetic before you dial
Never open a refi call with math you haven't already done. Here is a worked example. The numbers are illustrative; your file will differ.
Say a client funded $340,000 at 7.25% two years ago — $2,319 a month in principal and interest. The balance is around $333,200, and you can put them at 6.25% today with roughly $4,900 in costs rolled in.
Three things that panel does that a rate-alert text cannot.
- It states break-even plainly. Total cost divided by monthly savings. Twenty-one months on the 30-year option. If the client has mentioned selling next spring, you just talked yourself out of a loan, correctly.
- It shows the reset. A borrower two years into a 30-year note who takes a new 30-year note is buying back two years they already paid for. In this example the reset hands back about $80,000 of the interest savings that were sitting on the table.
- It puts the shorter term on the page. The 25-year saves less every month and far more over the life of the loan. It also pays you less today. Show it anyway — that is the whole reason they take your call in 2031.
Two more disciplines. A "no-closing-cost" refi carries its cost in the rate; that's a legitimate structure, and you say so out loud rather than letting them discover it. And if the client's real problem is PMI on a conventional loan, the answer may be an appraisal and a removal request, not a new loan at all. Saying that costs you one file and buys you a client.
Four gates before you dial
The first gate is the one branches skip. An early-payoff provision means a loan that pays off inside the window claws the compensation back from the branch — you funded the unit, worked the file, and ended up down on the month. Refinancing your own book straight through an EPO window is a very expensive way to look busy. Read your own agreement, know the window in months, and either wait it out or do the loan deliberately because the client's benefit is large and immediate. What you never do is pretend the window isn't there. If you can't say what one file actually earns the branch after a clawback, start with per-loan profitability and come back to this.
Gate four is the cheapest quality control in the business. Would you sign this with their numbers? If the honest answer is no, the call doesn't happen, and no volume target changes that.
Make it a monthly review, not a rate-drop scramble
A funded-book review that happens "when rates drop" is not a system. It's a scramble that arrives late.
- One list, five fields. Close date, note rate, product and term, MI status, current balance. If the LOS export won't produce it cleanly, build the sheet once and add a row at every closing. Five fields per funded loan is not a burden.
- Same day every month. Sort by rate gap, then run the calendar triggers regardless of where rates sit. An ARM reset doesn't care what your pipeline looks like.
- Every name has an owner. The originating LO owns their book. Files whose LO left go to a named person, not "the branch."
- Log every touch. An unlogged past-client call gets repeated or forgotten, and getting called twice in a month by the same branch reads as a campaign — which is precisely what you're trying not to be.
- Count the reviews that end in nothing. "Reviewed, no action" is a result worth tracking. Without it, the review quietly turns into a list of people you intend to pitch.
This is watch-work, and watch-work is what software should own. In MAVYN the funded book sits behind the same login as the pipeline and the branch P&L, and MAVIS watches for the triggers and drafts the outreach — which a human reviews and sends, every time. A refi conversation is advice. Advice doesn't go out on autopilot.
Open the call as a review, not a pitch
The framing is annual review. It's true, it repeats every year without getting stale, and it survives the client saying no.
The text, if you have a mobile number:
"[First name] — [your name] at [branch]. Once a year I check every loan I've closed against the current market. Yours came up this week. Two minutes on the phone and I'll tell you whether it's worth doing anything. It might not be."
The call opener, once they pick up:
"I pulled your file. You're at 7.25% on a 30-year, funded two years ago in March. Here's what today looks like, here's what it would cost, and then I'll tell you what I'd do if it were mine."
And the stay-put call, which is the one that actually compounds:
"I checked. Refinancing you right now costs more than it saves — you'd be about twenty-one months getting back to even, and you told me you might move. I'd sit tight. I'll look again in the spring."
Anyone who bought your client's balance can offer them a rate. Only the person who closed the loan can credibly tell them not to do it.
Work it like it has to last twenty years
Because it does. Churning runs the other way: serial refis that mostly generate fees will damage your early-payoff record first, then your investor relationships, then your referral flow — in that order, and faster than you think.
The funded book compounds only if every conversation leaves the client measurably better off. That's the entire trick. Two lanes of triggers a system watches for you, one lane you find because a human told you something, four gates before the dial, and honest arithmetic on the screen before you say hello.
Open the list this month. Sort it by rate gap. Call the top ten, and tell six of them to do nothing.