A first pass on a loan package, with the memorandum, rent roll and T-12 tied together and the loan sized, took Titleman's engine a median of 55.9 minutes in 43 finished runs. We have not timed a lender's own screen. A slow no costs the borrower days and the lender the analyst hours already spent.
The sections below show what was measured, what is arithmetic on a stated assumption, and how to count your own days.
What did we measure?
One query against production on 18 October 2026 at 09:03 UTC. It counts every underwriting run started in the previous 30 days that finished and was charged with credits, ours and our clients' alike. The clock runs from the start of the run to the delivery of its result.
| Measure | Underwriting runs |
|---|---|
| Finished and charged runs, 30 days to 18 October 2026 | 43 |
| Median, start to delivery | 55.9 minutes |
| The middle half of runs | 37.5 to 70.0 minutes |
| One run in ten took longer than | 93.4 minutes |
Where a run carries no delivery time, the time of its charge is used (5 of the 43). This is a full first pass, deeper than a screen, and we have not timed a screen alone. Do not read 55.9 minutes as the time to reject a loan. It is what happened, not a promise for your file.
What can the first pass say no to?
A no usually starts with documents that disagree. Two finished runs: a sample package on a QA account, and a package with the name and place removed and every money figure scaled by one factor, so ratios hold:
| Package | What the first pass found |
|---|---|
| 120 units, loan of 14,040,000 (sample package) | Restated as a hypothetical maturity of the 14,040,000 at 6.5 percent, the T-12 income gave 0.97x coverage and supported about USD 10,854,000 at 1.25x, which is USD 3,186,000 below that balance. The run itself read a purchase package. |
| 50 units, loft conversion | The memorandum's in-place NOI was 32 percent above the T-12. The rent roll and the T-12 agreed with each other. |
Both are one package each, not an average of anything. The working is in the NOI gap page and the loan-file check.
What does a slow no cost?
It falls on two parties. None of the figures below is a measurement of how often borrowers leave.
| Who | What it costs | What we know |
|---|---|---|
| Borrower | The days left before a maturity or a contract date | The dollar value depends on terms we do not hold. We print none. |
| Lender, analyst time | Analyst hours spent on files that end in no, times the loaded hourly cost | We hold neither number. Both are in your pipeline log. |
| Lender, fee | The origination fee on a file you would have said yes to but reached too late | Arithmetic: an assumed 1 percent of the 14,040,000 balance is 140,400 dollars, and of the 10,291,000 that the value test supports, 102,910 dollars. |
The fee is an assumption you can change. We have not measured how often a file waits behind a no.
How do you time your own screen?
- Log three dates per file: package received, first findings in the analyst's hands, answer sent.
- Take the median and the slowest tenth of each gap over thirty days.
- Split the files by outcome and look at how long the files that ended in no took.
- Run one closed file through an engine and compare its first pass with your own.
The result is a number nobody else can supply.
What does this not replace?
The first pass does not replace the credit decision, a loan approval or a term sheet, and it does not score the chance of default. It does not replace a licensed appraisal where a lender needs one to rely on a value, a title search, an engineer's inspection, or the call to the borrower about each finding. It does not cover documents still to arrive or a credit committee's calendar. The output is a draft for a licensed professional to review.
See it on a file you already declined
Request a demo and send a loan file you have already decided. We return the first pass and the open items, and you compare the findings and the time with your own file.
Prepared by Titleman AI, on behalf of the Titleman team.