Underwriting KPIs are the few numbers that show whether underwriters decide files quickly and correctly. Five cover most teams: turn time, files per underwriter, conditions per file, pull-through and defect rate. Track each by product and underwriter from loan system dates, and pair every speed measure with a quality measure so neither can be gamed.
Which underwriting KPIs matter most?
The underwriting KPIs that matter most are the ones that answer two questions together: how long files wait for a decision, and how often that decision holds up. Turn time and files per underwriter cover speed. Conditions per file, pull-through and defect rate cover quality and outcome. Five numbers are enough for a monthly review.
Each of the five tells you something the others cannot:
| KPI | Example definition | What it reveals |
|---|---|---|
| Turn time | Business days from complete submission to initial decision, median and 90th percentile | Queue and review speed |
| Files per underwriter | Initial decisions per underwriter per month, by product | Capacity and workload balance |
| Conditions per file | Average conditions on the initial approval | Submission quality and review clarity |
| Pull-through | Share of files submitted to underwriting that fund | Whether decisions lead to closed loans |
| Defect rate | Share of quality control reviewed files with a material finding | Whether decisions were correct |
Lenders add more, such as suspense rate at setup or condition rounds per file. Add them only when a specific problem calls for them. A dashboard of twenty numbers gets glanced at, and a list of five gets discussed.
Speed underwriting KPIs: turn time and files per underwriter
Speed underwriting KPIs measure how quickly decisions happen and how much each underwriter decides. Turn time is elapsed days between dated events, reported as a median and a high percentile. Files per underwriter is a count of decisions in a period. Both need cutting by product, because a bank statement file and a salaried refinance are different jobs.
Turn time
Report turn time from complete submission to initial decision as a median and a 90th percentile, with a separate clock for condition turn time, because an average hides the stuck files. Which clocks to run and how to cut them by stage and source is covered in how to measure underwriting turn times across the pipeline.
Files per underwriter
Count initial decisions, not touches. If a file comes back three times with conditions, it is still one decision, and counting each touch rewards slow condition work. Weight the count by product if the mix varies: one 24-month bank statement file may equal two or three standard files in effort. Set the weights from the team's own touch time, not from a figure borrowed from another lender.
Do not rank underwriters on raw volume. The person who takes the hardest files will always look slowest, and the incentive becomes avoiding them.
Condition count as an underwriting KPI
Condition count, measured as the average number of conditions on an initial approval, is an underwriting KPI because each condition is work for the borrower, the processor and the underwriter, and many are avoidable. A high count can mean weak submissions, a cautious underwriter or vague guidelines. The number starts the question rather than answering it.
Count fairly:
- Separate standard conditions from file-specific ones. Items every loan gets, such as a final title policy, inflate the count without saying anything about review quality.
- Track condition rounds too. A file with four conditions that clears in one round is in better shape than one with two conditions that bounce back three times.
- Cut by source. If one broker's files average twice the conditions of others, the fix is at intake.
Late-stage conditions deserve their own line. Items that appear only at the final checks before closing, such as an updated statement or a final employment confirmation, are often predictable. The guide to clear to close final checks lists what typically surfaces there. A verbal verification of employment is one example that can be scheduled rather than discovered.
Pull-through and defect rate as underwriting KPIs
Pull-through and defect rate are outcome underwriting KPIs: they show what happened to decisions after they were made. Pull-through is the share of underwritten files that fund. Defect rate is the share of quality control reviewed loans with a material finding. Speed means little if approved files fall apart before closing or fail review after it.
Pull-through
Pull-through falls for reasons underwriting does not control, such as rate moves or a borrower walking away, so read it with the reasons attached. Code each lost file: borrower withdrew, denied, appraisal issue, conditions not met, moved to another lender. The codes that underwriting influences, such as conditions not met and late denials, are the ones to track by underwriter.
A late denial is the worst result for everyone. A file approved with conditions and then denied after weeks of work usually means the problem was visible earlier. Track the share of denials that come after an initial approval.
Defect rate
Defect rate comes from post-closing quality control samples, or from pre-funding reviews where the lender runs them. Use the quality team's severity scale and count only material findings, such as income calculated wrongly or an unsourced deposit used for funds to close. Clerical misses go in a separate count.
Because samples are small, read defect rate over a rolling period of several months rather than month by month. One finding in a sample of ten looks like a ten percent rate and may mean little.
How do you track underwriting KPIs without gaming?
You track underwriting KPIs without gaming by pairing each speed measure with a quality measure, pulling every number from system dates rather than self-reports, and reviewing them with the team instead of only ranking people. A turn time target alone invites quick approvals with long condition lists. Paired with conditions per file, it does not.
Practical rules that keep the numbers honest:
- Pair them. Turn time with conditions per file. Files per underwriter with defect rate. Pull-through with late denial share.
- Use system timestamps. Submission, assignment, decision and clear-to-close dates from the loan origination system. Manual logs drift.
- Freeze definitions. If a definition changes, restate earlier months under the new one, or mark the break clearly.
- Report exclusions. If a file is left out of turn time, list it and say why.
- Discuss, then rank. A monthly meeting where the team looks at outliers teaches more than a league table.
A monthly underwriting KPI scorecard for one team
A monthly underwriting KPI scorecard puts the team figure, each underwriter's figure and the target on one page. The example below is for a fictional team of four underwriters. Every number is an example, made up to show how the figures interact, and is not a benchmark.
| KPI | Example target | Team, example month | Underwriter C, example month |
|---|---|---|---|
| Turn time, median (business days) | 3 | 3 | 2 |
| Turn time, 90th percentile | 8 | 9 | 5 |
| Files per underwriter, weighted | 45 | 47 | 58 |
| Conditions per file, file-specific | 5 | 5.4 | 8.1 |
| Condition rounds per file | 1.5 | 1.6 | 2.3 |
| Pull-through | 75% | 74% | 71% |
| Material defect rate, rolling 3 months | 3% | 2.5% | 4.0% |
Reading it: underwriter C is the fastest on the team and decides the most files. The quality side tells a different story. C issues half again as many conditions as the team, those conditions take more rounds, and C's defect rate is above target. The speed is partly borrowed from the condition stage. The conversation with C is about condition wording and statement review depth, not about slowing down.
The team's 90th percentile of 9 days, above its target of 8, would lead to a look at which files sat longest. If they are mostly bank statement files waiting in the queue, the issue is routing or capacity rather than any one person.
Underwriting KPIs for bank statement and MCA files
Underwriting KPIs for bank statement and MCA files need their own targets because the income or revenue is built from months of deposits rather than read from a document. Turn time runs longer, conditions run higher and the defects come from different places, mostly deposit totals, transfers and unsourced credits. Mixing these files into one blended target hides their problems.
For bank statement mortgage files, add two measures. The first is the share of files where the underwriter's income figure differs from the one submitted by more than a set percentage, which shows how well brokers and processors calculate income. The second is defects tied to deposits specifically: transfers counted as income, a missed month, a large deposit used without a source.
For business loans and merchant cash advances, the useful outcome measure is early performance: missed payments in the first months, by underwriter and source. Funders that sell parts of advances to other investors also need consistent underwriting records to share with them, and the guide to MCA syndication covers how those arrangements work.
Statement review time is often the biggest driver of turn time on these files. TrueBankFlow reads multi-month statements, totals deposits each month and ties every figure back to its statement line, which lets a reviewer check the math quickly when a quality finding needs tracing. To test it against your own scorecard, ask for a TrueBankFlow onboarding slot.