Transaction categorization for lenders gives every bank statement line one label that matters to a credit decision: payroll, business receipts, transfers, rent, loan payments, fees and a few others. The labels let a reviewer total income and spot obligations quickly. Transfers, cash, refunds and unlabeled credits still need an underwriter to confirm them by hand.
What category set does transaction categorization for lenders need?
A lender needs a short category list built around one question: does this line change what the borrower can repay? That usually means eight to fifteen labels, split between credits and debits. A bookkeeping chart of accounts has hundreds of lines, and almost none of them help an underwriter.
The difference is purpose. An accountant categorizes to produce a tax return and a profit and loss statement, so office supplies and fuel need separate homes. A lender categorizes to answer narrower questions. How much money comes in from earning activity? How much of it repeats? What does the borrower already owe each month? Which items look risky? Every label in the set should feed one of those answers.
A workable starting set looks like this:
- Credits: payroll, business receipts, government and retirement benefits, own-account transfers in, loan proceeds, refunds and reversals, cash deposits, other credits.
- Debits: rent or mortgage, loan and credit card payments, support payments, own-account transfers out, NSF and overdraft fees, cash withdrawals, other spending.
Two labels in that list do most of the work in practice: own-account transfers and "other". Transfers are the largest source of double-counted income. "Other" is where the categorizer admits it does not know, and a large "other" bucket is a sign the set or the descriptions need attention.
Keep the set stable across files. If one underwriter tags a client payment as "business receipts" and another as "sales", averages stop being comparable across the pipeline, and shared exclusion rules stop matching. A team that cannot agree on the list will find the disagreement shows up later as rework.
Income categories for statement credits
Income categories divide deposits by where the money came from, because the source decides whether it can count toward qualifying income. Payroll, business receipts and benefits are earning or recurring sources. Transfers, loan proceeds and refunds move money around without the borrower earning it, so they normally come out of the average.
Payroll
Payroll credits carry an employer or payroll processor name and arrive on a fixed schedule, weekly, every other week or twice a month. They are the easiest income to label and the easiest to check against a pay stub. A change in amount mid-period is worth a note, since it may be a raise, a cut in hours or a job change.
Business receipts
Business receipts are customer payments into a self-employed borrower's account: card processor settlements, ACH credits from clients, mobile deposits of checks and payment app transfers. They vary in size and timing, which is normal. What matters is that the payer is a customer rather than the borrower or a family member.
Benefits and other recurring sources
Social Security, pension, disability and similar deposits arrive monthly with a fixed amount and a government or plan name. They usually need an award letter to count, but the statement shows that the money actually lands.
Credits that are not income
Own-account transfers, loan proceeds, refunds, chargeback reversals and insurance settlements all raise the deposit total without adding earnings. Labeling them correctly is what keeps a total deposit figure from turning into an inflated income figure. The overview of what bank statement analysis gives a lender explains how these exclusions move the gap between total and qualifying averages.
Obligation categories in statement debits
Obligation categories mark the debits that represent a promise to pay each month: housing, loan and card payments, and support. They matter because they reveal debts the application or the credit report may not show, such as a private loan, a lease paid to an individual or a payment app installment plan.
Statement debits fall into three groups for a lender:
- Fixed obligations. Rent, mortgage payments, auto loans, student loans, personal loans, merchant cash advance debits and court-ordered support. These belong in the debt picture.
- Revolving payments. Credit card payments vary month to month. The statement shows the borrower paying, while the card balance and minimum payment come from elsewhere.
- Spending. Groceries, fuel, subscriptions and the rest. Lenders rarely need these broken out, so one "other spending" label is enough.
Daily or weekly debits to the same funder deserve their own label. A fixed ACH debit every business day often points to a merchant cash advance, and two of them suggest stacking. A monthly debit to a person, with a memo like "rent" or "car", may be an obligation the borrower forgot to list.
Support payments need care. A recurring debit to a state disbursement unit or an ex-spouse is an obligation that belongs in the ratios. Labeling it as "other spending" hides it.
Which transaction categories must an underwriter review by hand?
An underwriter should review every line labeled as a transfer, a cash deposit, a refund, a loan proceed or "other", plus any credit large enough to move the average. These are the labels where a wrong guess changes the income figure. Payroll and benefits labels can usually be checked in bulk against stubs and award letters.
A practical review order:
- Unlabeled and "other" credits. Each one is a question the file will get asked anyway.
- Own-account transfers. Confirm the other account belongs to the borrower. If the transfer partner is a business the borrower owns, the program rules decide whether it counts.
- Cash deposits. Cash is hard to source. Many programs want an explanation for regular cash, and a large single cash deposit usually needs a letter.
- Refunds and reversals. Confirm they are not customer payments with a misleading description.
- Large credits. Anything well above the borrower's normal deposit size, whatever its label.
- New debits to lenders. A payment that started recently may be a new loan not yet on the credit report.
This is also where a second reviewer earns their time. When a file depends on a few judgment calls about these categories, a second level underwriting review gives the calls a fresh look before the file moves on.
One month of statement lines with category labels
The table below is an example month for a fictional self-employed borrower. Every name and amount is made up. It shows the proposed category for each line and whether a person needs to confirm it before the line counts.
| Date | Description | Amount | Proposed category | Reviewer confirms? |
|---|---|---|---|---|
| 03-01 | ACH CREDIT HARBOR PLUMBING SUPPLY | 2,400.00 | Business receipts | No |
| 03-02 | ZELLE FROM M ORTEGA | 900.00 | Business receipts | Yes, payer could be family |
| 03-04 | ONLINE TRANSFER FROM SAV 7781 | 3,000.00 | Own-account transfer in | Yes, confirm account owner |
| 03-05 | RENT PAYMENT ZELLE TO J PARK | -1,850.00 | Rent | No |
| 03-08 | SQUARE INC DEPOSIT | 4,115.20 | Business receipts | No |
| 03-10 | ACME AUTO FINANCE PMT | -512.44 | Loan payment | No |
| 03-12 | MOBILE DEPOSIT | 6,200.00 | Other credit | Yes, no payer shown |
| 03-15 | CASH DEPOSIT BRANCH 214 | 1,500.00 | Cash deposit | Yes |
| 03-18 | AMAZON RETURN | 64.99 | Refund | No |
| 03-20 | NSF RETURNED ITEM FEE | -35.00 | NSF fee | No |
| 03-25 | ONLINE TRANSFER TO SAV 7781 | -2,000.00 | Own-account transfer out | No |
| 03-28 | STATE DISB UNIT CHILD SUPP | -640.00 | Support payment | No |
Read the right-hand column and four of the seven credits need a look. Their total is 11,600.00 of 18,180.19 in credits, so the review decides most of the month's income. If the 6,200.00 mobile deposit turns out to be a customer check, it counts. If it was the borrower's tax refund, it does not, and the month loses 6,200.00 of qualifying deposits in one decision.
The debit side also tells a story. Rent, an auto loan and child support add up to 3,002.44 a month in obligations. Only the auto loan is likely to appear on a credit report. The NSF fee is a single event in this month, but it belongs in the count across the whole lookback.
Categorization errors on transfers and refunds
The most costly categorization error labels money the borrower moved as money the borrower earned. A transfer from savings tagged as business receipts, or a reversal tagged as a customer payment, inflates the deposit average without any warning. The file looks stronger than it is until someone traces the line.
These errors have recognizable causes:
- Vague descriptions. "DEPOSIT", "ONLINE TRANSFER" or "MOBILE DEPOSIT" with no payer give a categorizer nothing to work with. A safe default leaves the line in "other" for a person, rather than guessing income.
- Accounts not in the file. A transfer can only be matched when both statements are present. When the savings account was never collected, the incoming transfer looks like an outside deposit.
- Payment apps. A Zelle or Venmo credit from a named person could be a customer, a roommate paying rent or a relative. The name alone does not settle it.
- Round trips. Money that leaves for another account and comes back a few days later should not count twice. Matching amounts near each other in time is the clue.
- Business-to-personal moves. An owner's draw from the business account into a personal account is a transfer between two accounts, even though it is how the owner pays themselves. Programs differ on which account's deposits they use, and counting both doubles the income.
The guard against all of them is the same. Anything a categorizer is unsure about should stay visibly unsure, so a person sees it before it moves an average.
Teams that build their own categorizer meet these errors early and often. The comparison of building or buying bank statement analysis looks at what it takes to maintain that logic as bank descriptions change.
Transaction categories in the loan file
Transaction categories reach the loan file as totals and as evidence. The report shows monthly totals by category, the qualifying deposits after exclusions, and the lines behind each excluded or flagged item. An underwriter or auditor should be able to pick any total and see which lines built it.
Three habits keep categories useful after the analysis ends:
- Record the overrides. When a reviewer changes a label, the file should show the original label, the new one and a short reason. "Moved 03-12 mobile deposit to business receipts: check image shows client name" is enough.
- Keep the category names stable. Downstream fields, conditions and reports depend on the same label meaning the same thing on every file.
- Pass totals, not just a PDF. When the analysis feeds an origination system, structured fields such as qualifying income, NSF count and recurring obligations save the processor from retyping. What to ask a vendor about those fields is covered in the guide to loan origination system integration for statement data.
TrueBankFlow sorts each transaction into income, payroll, rent, loan payments and other categories that the reviewer can edit, and every total links back to its statement lines. Teams can request an early access spot to try it on their own files.