The closing is Thursday at two. Wednesday at 4:40, your processor notices the payoff on the borrower's departing residence was ordered a week ago and never came back. She calls. The payoff department's phone queue closed at 4:30 Central. Thursday morning she reaches a human, who explains the request went to a retired fax line and standard turn is two business days. The closing moves to Monday. The listing agent calls you. The borrower already booked movers.
Nobody was careless. Your processor caught it the second she looked. The failure happened at 8:10 Wednesday morning, when that file already read payoff ordered — day six, no response and nobody's eyes were on the list where that sentence lives.
The short version
- Your morning is a race against other people's intake cutoffs — title, payoff departments, HOA management, the appraisal desk, the lock desk. A problem found at 8:10 costs nothing. The same problem at 4:40 costs a business day.
- Sort exceptions by dependency depth, not close date. A three-party chain nine days out outranks a one-call problem closing tomorrow.
- Scan for asks you are waiting on, not for status. Any file whose most recent event is "we asked somebody" and is past turn time gets a call today.
- Every ask carries a name, an hour, and an out: tell me now if it won't be back by noon.
- Score the morning by asks that left the building before nine — not by tabs closed.
Your morning is a cutoff race, not a status check
Almost nothing that kills a closing lives inside your building. Title, the payoff department, HOA management, the appraisal desk, the VOE vendor, the investor's lock desk, the underwriting queue — every one of them runs a business day with an intake cutoff and a turn time measured in days.
That changes what a delay costs. Find a missing payoff at 8:10 and you're in today's queue; the request works while you eat lunch. Find it at 4:40 and you're in tomorrow's queue, and if tomorrow is Friday you're in Monday's. The unit of loss is not minutes. It is business days, and nobody gives partial credit for a request sent at 4:41.
So the morning question is not how is the pipeline doing. It is: what has to leave this building in the next fifty minutes to land inside somebody else's today?
Sort by dependency depth, not by close date
Most managers scan the pipeline by closing date. It feels responsible and it is the wrong order, because closing date tells you when the pain arrives, not how long the cure takes.
Ask a different question of every exception: how many separate parties have to act, in sequence, before this file is clean?
- Depth 0 — you or the LO fix it with one call or one document. A letter of explanation, a missing paystub, a signature.
- Depth 1 — one outside party, one turn time. Order the appraisal. Request the payoff. Send the condition into underwriting.
- Depth 2 or more — chained. Payoff comes back, then title rebuilds the settlement statement, then the disclosure goes back out, then the borrower's waiting period runs. Four links, four calendars, and no link starts until the one before it finishes.
Which gives the sentence that reorders a morning: a depth-2 item nine days out beats a depth-0 item closing tomorrow. The depth-0 item has a fix that fits in an afternoon. The depth-2 item needs four business days it has not started spending, and no amount of urgency on Wednesday manufactures Monday.
The twenty minutes, in the order that pays
- Minutes 1–4: today and tomorrow, read backwards. Not "is it clear to close." Start at the wire and walk back — funding conditions, settlement statement, payoff figures, disclosure timing. The only thing you're hunting is the last item that still has to arrive, and whose building it's sitting in.
- Minutes 5–9: the outbound queue. Every file whose most recent event is we asked somebody for something. Ordered, requested, submitted, sent to the borrower. Anything past its normal turn time earns a call today. This is the highest-yield stretch of a manager's day and virtually nobody schedules it, because it isn't a report anyone ships.
- Minutes 10–13: locks inside ten days, then overnight leads. Locks below. Leads are the one thing you act on mid-routine instead of noting — an unassigned lead from 9 p.m. gets a name right now, because that clock started twelve hours ago.
- Minutes 14–17: yesterday by person, plus capacity. Contacts, applications, follow-ups worked. One slow day is a slow day; two or three in a row is a check-in, not a confrontation. And read your processor's load while you're here — an LO's slow week is visible, but a processor two files past capacity is invisible until files start stalling, and then you find it in step two, a week late and a business day more expensive.
- Minutes 18–20: three priorities, written down. Each has to move money, unblock a person, or prevent a specific problem you just spotted. "Work on recruiting" doesn't qualify. "Call the closer about Thursday's payoff before ten" does.
Steps 1 and 4 produce notes. Step 2 produces asks. If you only ever protect one block, protect that one. This is also the part software can genuinely carry: in MAVYN the morning is composed before you sit down — one login for pipeline, leads, coaching, recruiting and branch P&L — with MAVIS watching files for stalls so the outbound queue is already sorted when you open it. The scan is still yours. The asks are still yours.
The lock math you do at 8:10, not on day zero
Say a $420,000 loan, and an investor that prices extensions at three basis points a day. That's $126 a day. A ten-day extension is $1,260 on one file.
Now the part that decides who cares. You can't take it out of the LO's compensation — comp plans in this business don't flex file by file. So the extension is a branch expense, landing squarely on that loan's per-loan profitability. Which is exactly why the person watching lock expirations at 8:10 should be the person whose P&L absorbs them, not the person carrying the file.
A lock at day ten is a conversation: chase the two open conditions, ask the agent to move the signing up three days, decide whether a short extension or a re-lock is cheaper. The same lock at day zero is a purchase. You pay the desk's number and tell the borrower afterward.
Every ask needs a name, an hour, and an out
"Any update?" is the most expensive question in mortgage operations. It hands the work back to you, it invites still working on it, and it buys a business day of nothing. Every ask that leaves your desk before nine carries three things instead: a named person, a specific hour, and explicit permission to give you bad news early.
To the closer on a stalled payoff:
"[Address], closing Thursday. Payoff was ordered last week and hasn't come back. I need figures by noon today. If that's not going to happen, tell me now so I start the workaround instead of waiting on it."
To an LO on a file that hasn't moved in three days:
"Nothing's moved on this since Thursday. I'm not asking what happened. Who's next to touch it, and what do they need from us today?"
To a listing agent when a date is genuinely at risk, before they hear it anywhere else:
"Wanted you to hear this from me while it's still fixable. We're waiting on one item. I'll know by two whether Thursday holds. If it doesn't, you'll get a new date and a reason from me, not an apology."
The pattern underneath all three: you are never asking for status. You are asking for a specific artifact, by a specific hour, with an open invitation to tell you no while no is still cheap.
What the routine is allowed to skip
- Email. Timeboxed twice — after the routine, and mid-afternoon. Never before. Email is other people's priorities sorted by recency.
- Reports and the P&L. One real sitting a week beats a nervous glance every morning.
- Vendor and account-rep calls. They'll take every morning you hand them.
- Recruiting outreach. A protected afternoon block. It matters too much to do in scraps.
- Non-urgent approvals. Batch them.
The test for anything trying to jump the line: can it be handed to somebody else's queue before their cutoff today? If not, it isn't morning work, no matter how loud it is.
Score it by what left the building
Not tabs closed. Not files reviewed. Count the asks that went out before nine, each with a name and an hour on it. Three is a real morning. Zero means you read the pipeline instead of working it.
Then run the audit once, honestly. Pull every closing date that moved last quarter and ask, file by file, how many were visible three mornings earlier on a list nobody was reading.
Every blown closing has a morning where it was still cheap.
Twenty minutes. Read the branch before the inbox, sort by depth instead of date, and leave the desk having put three things into other people's hands. The routine doesn't make the work smaller. It moves discovery earlier, which is the only place the discount lives.