How consistent are document naming conventions across a mortgage processing team?

A file with names like scan0012, statement-final, and borrower-docs-2 forces the next person to open every item to learn what it is. Inconsistent naming slows review and increases the chance that a document is missed. This protocol measures how closely a team follows its own naming rule, then separates the misses caused by a confusing rule from the misses caused by inattention. The measure starts with the rule the team says it follows, then tests the names against it.

Question and scope

The question is how consistently document filenames on a defined set of mortgage files follow the team's approved naming convention. The scope names the loan types, the document categories, the storage location, and the naming rule version. A team might study income and asset documents on purchase files in one processing pod. The result describes that setting and should not be generalized to every document or every office.

Write the naming rule before collecting data. The rule should name the required segments, the order, the allowed separators, and the date format. It should also say how to handle a document that does not fit, such as a combined statement. A rule that only says use a clear name cannot be measured because two reviewers will disagree about clarity.

Freeze the rule version for the observation window. If the rule changes during the study, record the date and analyze the periods separately. A moving rule makes every inconsistency ambiguous. The version is part of the comparison, not a background detail.

Unit of analysis

The unit is one file on one document. A file with twenty documents contributes twenty units, but the report should also show the file-level result. That prevents a file with many small documents from dominating the analysis. Record the file key, the document type, the filename, the storage location, the upload time, and the uploader.

Record the uploader as the person who saved the document, not the person who requested it. The uploader is the one who applied the naming rule at that moment. If the system does not capture the uploader, the study has a data limitation worth documenting. A shared account makes attribution impossible and should be reported as such.

Keep borrower identity out of the working dataset even when the filename contains it. Use a file key that an authorized reviewer can link back to the system. The loan processing support lead should approve the fields and the masking rule. Filenames can contain personal information, so the study file should store a redacted version where possible.

Defining consistency

A consistent name follows every required segment in the right order and uses the approved separators and date format. A partially consistent name has the right information but in the wrong order or with a nonstandard separator. An inconsistent name lacks a required segment or uses a label that cannot be understood without opening the file. A name that is consistent but unreadable still fails the practical test.

Score the presence and order of segments separately. A name can include the borrower, document type, and period but put the period first. That is a format error, not a missing-information error, and the fix is different. A name that omits the period is a missing-information error that may require retraining or a better template.

Add a category for names generated by the system. Some portals assign a long identifier that the uploader cannot change. Those names should be scored separately from names the team controls. Otherwise the consistency rate will reflect the portal rather than the team's habits. The category also helps the team decide whether a system setting needs to change.

Sampling and observation

Use consecutive files from a defined window, or a random sample with a recorded start. Include documents from each category and each uploader if the volume allows. If the sample includes only the most recent files, it may reflect a recent training push rather than the normal process. Record the sample dates and any known training event.

Record the number of files excluded and the reason. A file with no documents cannot be scored, and a file stored outside the approved location may not follow the same rule. Report the exclusions beside the rate. A high exclusion count is a finding about storage discipline, not only about naming.

Capture the filename exactly as stored, including spaces, capitalization, and separators. Do not normalize the name before scoring, because the normalization is part of what the rule requires. If the analysis tool changes the name, keep the original in the record. The exact string is the evidence.

Data quality checks

Check that every row has a filename, a document type, and a storage location. A row with a missing field cannot be scored. Mark those rows separately rather than treating them as consistent. A high missing-field rate may point to a storage location that is not being used as intended.

Have a second reviewer score a subsample and report agreement by consistency category. Reviewers may agree on clearly consistent names and disagree on partially consistent ones. The category-level result shows where the rule needs an example. Keep one disagreement for each category so the team can discuss the standard.

Check for duplicate names within the same file. Two documents with the same name create a different problem from a nonstandard name, because one may overwrite the other. Report duplicates separately and check whether the storage system adds a suffix. A suffix can hide the collision without resolving the confusion.

Analysis and reporting

Report the number of documents, the consistency rate, and the share by category. Add a file-level view showing how many files had at least one inconsistent name. A file with one partial name is different from a file where every name is nonstandard. The file view helps the team set a realistic target.

Break the rate down by document type and uploader. One document type may have a long, awkward name that encourages shortcuts, while another may be easy to name consistently. One uploader may have learned an older rule. The breakdown points to a template fix or a coaching conversation rather than a general reminder.

Compare naming consistency with the time to find a document during review. A consistent name should reduce search time, but the study can test that assumption on a small sample. If the time does not improve, the rule may be too complex or the search may rely on another field. The comparison keeps the naming rule tied to a practical outcome.

Limits and sources of error

The study measures adherence to the team's rule, not whether the rule is good. A team can follow a confusing rule consistently and still create slow reviews. Include a short review of the rule itself and ask whether a new user can apply it without help. A consistent but unusable rule is still a problem.

The sample may exclude documents that were never uploaded or were stored in a personal drive. Those documents do not appear in the naming data, even though they affect the file. Add a count of documents found outside the approved location during a file review. That count gives the reader a sense of the work the study cannot see.

The uploader field may be unreliable when the team uses shared accounts or bulk imports. Report the share of rows with a known uploader and avoid drawing conclusions about individuals when the field is uncertain. The team-level result is more robust than an uploader-level result built on weak attribution. A team-level result is more useful than an individual score when the uploader field is unreliable.

Review cadence

Repeat the protocol after a naming rule change, a new portal, or a training session. Use the same categories and sampling rule so the comparison is meaningful. If consistency rises but search time does not fall, the rule may have become more rigid without becoming more useful. Report both movements.

Share the result with the people who name the documents. Ask which document types are hardest to name and why. The mortgage operations support lead can decide whether the fix is a shorter template, an example sheet, or a system default. A rule that people can apply in seconds is more likely to hold.

End the report with the document type that remains least consistent and the evidence needed to fix it. Name the owner and the date. The goal is a file where the next reviewer can identify a document from its name and spend attention on the evidence rather than the search. The next reviewer should be able to identify the document without opening it.