Do mortgage document checklists match the conditions actually requested?

A checklist can look complete and still miss the document a processor requests on a real file. The reverse also happens, where the checklist asks for an item the condition no longer requires. Both create wasted requests and confused borrowers. This protocol compares the checklist a team publishes with the conditions its processors actually write, then classifies the differences so the team can fix the source.

Question and scope

The question is how closely the team's document checklist matches the conditions requested on a defined set of mortgage files. The scope names the loan types, the checklist version, the processing stage, and the condition categories. A team might compare its purchase-file checklist with the conditions requested between application and underwriting over one month. The result describes that setting and should not be generalized to every program.

Define what counts as a checklist item and what counts as a condition. A checklist item is a document the team expects for a file type or stage. A condition is a specific request written on a file for a specific borrower. The two can share a name and still differ in period, signer, or acceptable source. The protocol should compare the substance, not only the label.

Freeze the checklist version for the observation window. If the checklist changes during the study, record the date and analyze the periods separately. A moving checklist makes every comparison ambiguous, because a difference may reflect a version change rather than a process gap. The version is part of the unit of analysis.

Unit of analysis

The unit is one checklist item matched to one condition category on one file. A checklist item may match several conditions, and a condition may require several documents. Record the file key, the checklist item, the condition category, the document requested, and the match result. The file key lets the analysis show how many differences appear on the same loan.

Use a matching table with one row per item and condition. A checklist item that appears on every file should be compared with the conditions that appear on the same files, not with a global list. That keeps the comparison tied to the work. A global list can hide a mismatch that only appears in one loan type.

Keep borrower identity out of the working dataset. Use file and condition keys that an authorized reviewer can link back to the source. The mortgage operations support lead should approve the fields and the review process. Checklist data can reveal training gaps, so the protocol should state how the findings will be used.

Defining match and drift

A match means the checklist item and the condition ask for the same document, period, and signer, with no material difference in the acceptable source. A missing item means the condition asks for something the checklist does not include. An extra item means the checklist includes something the condition does not require. A mismatch means both exist but differ in period, signer, or source.

Add a category for an unclear condition. An unclear condition may not match any checklist item because the processor did not specify enough detail. That is different from a checklist gap, and the fix belongs in the condition template. Keeping the category separate prevents the team from adding checklist items to compensate for vague requests.

Do not score a match from the document name alone. A pay stub checklist item and a pay stub condition can differ on the number of pay periods or the acceptable employer format. The reviewer should read the condition and the checklist note before deciding. The match table should record the specific fields compared.

Sampling and observation

Use consecutive files from the defined window, or a random sample with a recorded start. Include files from each loan type and each processor if the volume allows. If one processor writes most of the conditions in the sample, the result describes that person's habits rather than the team's checklist. Report the processor count so the reader can judge the coverage.

Record the checklist version used on each file. A file processed under an older version should be scored against that version, not the current one. Mixing versions creates false drift. If the team cannot tell which version applied, mark the file as unknown and report the count. A high unknown count is a version control problem.

Capture the condition text and the checklist text for every match. The analysis needs the exact wording to classify a mismatch. A summary written after the fact may smooth over the difference the study is trying to find. Store the text in a way that authorized reviewers can compare without exposing borrower data.

Data quality checks

Check that every row has a checklist item, a condition category, and a match result. Rows with a missing field cannot be scored. Mark them separately rather than treating them as matches. A high missing-field rate weakens the study and may point to a checklist that is hard to apply.

Have a second reviewer classify a subsample and report agreement by drift category. Reviewers may agree on the overall match rate and disagree on whether a difference is material. The category-level result shows where the definition needs work. Keep one example of each disagreement for the team discussion.

Check the checklist distribution. If the checklist is stored in several places with different dates, the team may be comparing against the wrong version. Record the source and the date of the copy used. The loan processing support workflow should name one authoritative checklist for each loan type before the study starts.

Analysis and reporting

Report the number of items compared, the match rate, and the share of missing, extra, mismatched, and unclear items. Add a file-level view showing how many files had at least one missing or mismatched item. A file with one extra item is different from a file with five missing items, and the team needs both views to set priorities. A file with one extra item is different from a file with five missing items, and the two need different fixes.

Compare the drift rate by document category and loan type. Income documents may match well while asset documents drift, or one loan program may have a checklist that no longer reflects the conditions. The pattern tells the team where to start. A single overall match rate hides those differences.

Report the source of the condition when it can be identified, such as a lender overlay or an investor guide. A missing item that comes from a new overlay may need a checklist update, while a mismatch that comes from a processor's habit may need coaching. The source field makes the fix more precise. The Fannie Mae Selling Guide and the Freddie Mac Single-Family Seller/Servicer Guide are common references for these overlays.

Limits and sources of error

The study measures the checklist against the conditions in the sample, not against every possible condition. A checklist can match the sample and still miss a rare document. Report the sample coverage and avoid claiming that the checklist is complete. The result is evidence about the observed files, not a certification.

The reviewer's judgment affects the match classification. A strict reviewer may call a small wording difference a mismatch, while a lenient reviewer may call it a match. The second-reviewer check estimates that variation, but it does not eliminate it. Report the agreement rate and describe the rule used for close calls.

Conditions can change after the observation window because a lender issues a new guideline. The protocol should record the date of the guideline and avoid treating every later change as drift. A checklist update that follows a guideline change is a normal response, not a defect. The next study should be designed around the new version.

Review cadence

Repeat the comparison after a checklist update, a new lender overlay, or a condition template change. Use the same categories so the comparison is meaningful. If the match rate rises but the unclear category also rises, the team may have made the checklist broader without making the conditions clearer. Report both movements.

Share the missing and mismatched items with the people who maintain the checklist and the condition templates. Assign each category to one owner and a date. The mortgage operations support lead can decide whether the fix is a checklist edit, a template change, or a training note. A finding without an owner becomes the next study's finding.

End the report with the document category that still cannot be matched consistently and the evidence needed to resolve it. Name the owner and the date. The goal is a checklist that reflects the conditions the team actually writes, so a processor can use it without translating or guessing. A checklist that cannot be matched to a real condition should not be treated as complete.