Skip to content

Guide · Mortgage and lending operations

Mortgage loan document capture: split, index and check the file

The documents in a loan file and the fields worth capturing from each, how to read a Closing Disclosure, cross-checks between documents, the traps in stacked files, and a step-by-step setup.

Practical guide · about 9 minutes to read · updated October 2026

Why loan files are hard to index

A loan file is dozens of documents from a dozen sources: the borrower's application, their employer's pay stubs and W-2s, their bank's statements, the appraiser's report, the lender's own disclosures. Processors, underwriters, closers and post-closing all need to find the right document fast, and every one of them has to belong to the right loan.

The pain is in the stack. Documents run from one page to dozens, statements repeat their header on every page, and the documents that matter most for income and assets never print your loan number at all.

The documents and their fields

The Mortgage sample job that ships with CapturePoint 6 recognizes these documents and reads these fields. Most teams add a few more, shown on the right.

DocumentLoan application
Fields the Mortgage sample job readsLoan number, borrower name, lender
Often addedCo-borrower, loan amount, property address
DocumentLoan Estimate
Fields the Mortgage sample job readsLender, borrower, loan ID, interest rate, estimated closing costs, estimated cash to close
Often addedDate issued, loan amount
DocumentClosing Disclosure
Fields the Mortgage sample job readsLender, borrower, loan ID, interest rate, cash to close
Often addedDate issued, closing date, loan amount, closing costs
DocumentPay statement
Fields the Mortgage sample job readsEmployee, employer, pay date, gross pay, net pay
Often addedPay period, year-to-date gross
DocumentW-2
Fields the Mortgage sample job readsEmployee, employer, wages, federal income tax withheld, tax year
Often addedEmployer EIN (never in file names)
DocumentVerification of employment
Fields the Mortgage sample job readsEmployer name, requesting organization
Often addedHire date, position, date completed
DocumentBank or account statement
Fields the Mortgage sample job readsStatement issuer, account holder, closing balance
Often addedStatement period, last four digits of the account
DocumentAppraisal report
Fields the Mortgage sample job readsParcel number, client
Often addedProperty address, appraised value, effective date

Reading the Loan Estimate and Closing Disclosure

The Closing Disclosure is the document everyone checks. Its first page holds most of what you need, in a layout set by regulation, which makes it one of the more predictable documents in the file.

Field map: Closing DisclosureA schematic first page of a Closing Disclosure: closing, transaction and loan information across the top, the loan terms table, projected payments, and costs at closing with cash to close at the bottom.CLOSING DISCLOSURE1234
  1. Closing, transaction and loan informationDate issued, closing date, borrower, lender, property, loan term, purpose, product, loan ID.
  2. Loan termsLoan amount, interest rate, monthly principal and interest, and whether each can increase.
  3. Projected paymentsPayments over time, with estimated taxes, insurance and escrow.
  4. Costs at closingClosing costs and cash to close.
A schematic of page one, not the official form. The Loan Estimate uses a similar first page, which makes the two easy to compare.

A worked example: the numbers check each other

This is a fictional Loan Estimate from the mortgage sample documents that come with CapturePoint 6: a 30-year fixed-rate purchase loan. The sample job reads the loan ID, interest rate, estimated closing costs and estimated cash to close; the rest is printed on its first two pages.

Value on the Loan EstimateLoan ID
Amount1895588776
Value on the Loan EstimateLoan amount
Amount281,000.00
Value on the Loan EstimateInterest rate
Amount6.887%
Value on the Loan EstimateMonthly principal and interest
Amount1,848.22
Value on the Loan EstimateD. Total loan costs
Amount4,027.00
Value on the Loan EstimateI. Total other costs
Amount6,155.00
Value on the Loan EstimateJ. Total closing costs (D + I)
Amount10,182.00
Value on the Loan EstimateDown payment
Amount31,000.00
Value on the Loan EstimateDeposit
Amount500.00
Value on the Loan EstimateEstimated cash to close
Amount40,682.00

Every line can be checked from the others. Loan costs plus other costs give the closing costs: 4,027.00 + 6,155.00 = 10,182.00. Closing costs plus the down payment, less the deposit already paid, give the cash to close: 10,182.00 + 31,000.00 - 500.00 = 40,682.00. And 281,000 over 30 years at 6.887% comes to a monthly payment of 1,848.22. A captured value that breaks one of these sums was misread, or the document has a problem, and either way a person should look before the file moves on.

Cross-checks worth automating

Capturing values from several documents in the same file lets you compare them. These comparisons catch real problems, and they are cheap once the data exists:

CompareInterest rate on the Closing Disclosure
WithThe latest Loan Estimate
What a difference meansA changed rate or a missing revised Loan Estimate.
CompareCash to close on the Closing Disclosure
WithEstimated cash to close on the Loan Estimate
What a difference meansCosts moved; someone should know why before closing.
CompareBorrower name on every document
WithThe application
What a difference meansA document filed to the wrong loan, or a co-borrower to add.
CompareWages on the W-2
WithYear-to-date gross on the latest pay stub
What a difference meansIncome that changed, or a pay stub from the wrong year.
CompareAccount holder on bank statements
WithBorrower and co-borrower names
What a difference meansAssets that may not belong to a borrower.

The rules themselves belong in your loan origination system or a simple spreadsheet. Capture's job is to put clean, typed values for every document in one place.

Common traps

Documents with no loan number

Pay stubs, W-2s, bank statements and verifications come from third parties and know nothing about your loan number. Either scan each loan file behind a barcode cover sheet, or look the borrower up in your pipeline and take the loan number from the match.

Statements that look like new documents on every page

A bank statement repeats the bank's header and account block on every page, so page three looks like page one of the next statement. Test separation on your longest statements, or use separator sheets for the files where a merged or torn document is unacceptable.

Several versions of the same document

A loan can have more than one Loan Estimate and more than one Closing Disclosure. Capture the date issued, and name files by it, so the latest version is obvious.

Private numbers

Applications and W-2s carry Social Security numbers, and statements carry account numbers. Keep them out of file and folder names. In CapturePoint 6's naming window, fields named as an SSN, tax ID or bank account are marked private, and using one in a name brings up a warning.

Copies, faxes and handwriting

Faxed verifications and photocopied statements read less cleanly, so expect them in review more often. Handwritten values are not something to count on.

Setting up capture, step by step

  1. Run the Mortgage sample. Install CapturePoint 6 (the free trial starts on first launch, with no form), pick Mortgage on the welcome screen and scan the sample loan files with the built-in demo scanner.
  2. Teach it your files. Choose Use my own documents and add a folder of real closed loan files, including your longest statements and appraisals. Setup finds where each document starts and ends, sorts them into types and proposes fields.
  3. Decide how documents start. Automatically, at barcodes, at both, or only at separator sheets. Barcode cover sheets per loan also give every page the loan number.
  4. Add your pipeline. Import a CSV export of active loans (loan number, borrowers, property) and look up the borrower name, so third-party documents get the loan number.
  5. Review. Needs review shows each waiting document with the reason. Confirm with Ctrl+Enter; corrections teach the job.
  6. Name and file by loan. For example Loans / 2026-004817 / Closing Disclosure_2026-03-14.pdf.
  7. Export. To Ademero Content Central, or a folder with a searchable PDF, a data file with every value and a text file. Each PDF can also go to SharePoint or OneDrive (work or school accounts), Google Drive, Dropbox or Nucleus One.
The CapturePoint 6 welcome: pick Mortgage to open the ready-made mortgage sample job.

Where the data goes next

  • Into a loan packet. In Content Central, a packet template keyed on the loan number gathers every document that carries it, marks each document type required or optional, and shows Complete or Incomplete with the missing documents listed. The Incomplete Packets report can run on a schedule. See how packets work.
  • Through review as one file. A Content Central workflow can route a whole packet for approval, and merge the packet's documents into one file when it is time to deliver it.
  • Into the loan origination system. The data file carries the values for each document, ready to import or compare.
  • Into retention. Content Central keeps each document type on its own retention schedule, with legal holds when needed.
Once you confirm a few documents, CapturePoint 6 re-checks the ones still waiting and clears those that now pass your thresholds (shown on the Invoices sample job).

Checklist

Before loan file capture goes live

  • Document types set up from real closed files, not one perfect example.
  • Loan number on every document, from cover sheets or a pipeline lookup.
  • Separation tested on long statements and appraisals.
  • Date issued captured on Loan Estimates and Closing Disclosures.
  • Cross-checks agreed: rate, cash to close, names, income.
  • No SSN or account number in any file or folder name.
  • Packet template with required documents, and a schedule for the incomplete report.
  • Retention set per document type.

Loan file questions

Does CapturePoint 6 check TRID timing or tolerances?

No. It captures the values (rates, closing costs, cash to close, dates) so your loan origination system or your team can compare them. Compliance decisions stay with the people and systems responsible for them.

How do pay stubs and bank statements get the loan number?

They do not carry one, so it has to come from somewhere else. Either scan each loan behind a barcode cover sheet, or look the borrower name up in an export of your pipeline and fill the loan number from the matching row. A name that fits two loans waits for a person.

Can it read handwritten applications or wet-signed pages?

Do not count on handwriting. Typed and printed documents are what to plan around; handwritten values should be confirmed by a person. Capture software also does not verify signatures.

Can we use it for files that arrive as PDFs from brokers?

Yes. CapturePoint 6 imports PDF, TIFF, JPEG and PNG files as well as scanning paper. If the documents should flow into another system as data, Paige, Ademero's cloud service, reads PDFs and returns the data by download, SFTP or webhook.

Next step

Scan a sample loan file.

Download CapturePoint 6, pick Mortgage on the welcome screen and watch a stacked loan file split into applications, disclosures, pay stubs and statements on your own PC. Want packets shown in Content Central? Book a demo.