Bank statements for a business loan: what the lender is checking

A lender has asked for twelve months of business bank statements, and unlike a tax return or a set of accounts, nobody prepared these for the purpose. They are the raw record, which is a large part of why they are asked for. Most of what gets decided from them comes out of a small number of measures computed off the transaction rows.

Those measures are computable by you, from the same rows, before you submit. Thresholds and product rules vary a great deal between lenders, so treat what follows as the underlying concern rather than a standard you can qualify against.

Average daily balance

Where a lender publishes a minimum balance requirement, it is commonly expressed as an average daily balance, and that is not the average of your month-end balances. It is the closing balance on every day of the period, summed and divided by the number of days.

Take a thirty day month where the account holds 12,000.00 for ten days after a large customer payment, drops to 4,000.00 for fifteen days, then sits at 800.00 for the final five. The daily balances total 120,000.00 plus 60,000.00 plus 4,000.00, which is 184,000.00. Divided by thirty, the average daily balance is 6,133.33.

The advantage of the measure is that it cannot be improved by timing. A closing balance can be lifted by holding a payment until the last day of the period; a daily average reflects the money that was in the account throughout. Compute it for each of your twelve months and read the twelve figures as a series rather than as one number, because a series that falls month on month reads very differently from a flat one at the same average, whatever the threshold turns out to be.

Deposit consistency, and what is actually a deposit

The second measure is whether revenue is steady. Both the total deposited each month and the number of deposits carry information here, since ten customer payments a month describes a more resilient business than one payment of the same total, and criteria that mention deposit consistency often refer to both.

The trap is that the credit column contains more than revenue. Consider a year with 240,000.00 of total credits. Inside that figure are 45,000.00 of owner contributions from a personal account and a 15,000.00 drawdown from an existing facility. Genuine revenue is 240,000.00 less 60,000.00, which is 180,000.00, or 15,000.00 a month rather than the 20,000.00 the raw total suggests.

Expect that adjustment to be made whether or not you make it. Making it yourself, and labelling the transfers so they are recognisable, is the difference between a file that reads as prepared and one that reads as optimistic. Where money moves between accounts you control, the reliable method is matching amount and date across the accounts, which is how transfers between accounts are traced rather than guessed from descriptions.

Negative days and returned items

The third measure is a count: how many days the account closed below zero, and how many payments were returned unpaid. Both bear directly on whether the business can carry another fixed repayment, and both are cheap to count yourself and unpleasant to be told.

Recency generally counts for more than the raw total. Three overdrawn days in month two of the run and three in the month before the application describe different businesses. Find them yourself by sorting the combined rows by date and reading the balance column for anything below zero, then look at what was happening around each one.

Returned items are found in the description column rather than the balance column. Filter for the bank's own wording for a returned item, a reversal or an insufficient funds fee, and treat each as something to be ready to explain in one factual sentence.

Existing debt service

The fourth measure is money already leaving the account to another lender. The pattern is distinctive: a fixed amount debited every business day, or every week, to a finance company.

A debit of 210.00 on each of 21 business days is 4,410.00 a month, and it comes out before payroll, rent or stock. A business showing 15,000.00 of monthly revenue with 4,410.00 of existing daily debits has a very different capacity from one showing 15,000.00 with none. An obligation of this shape appears on the statements whether or not it appears in the accounts, which is a large part of why statements get asked for at all.

Group the debit column by counterparty and by amount to surface it. What you are looking for is an identical amount leaving on a regular cadence to a counterparty that is not a supplier. Frequency on its own does not identify anything: a daily card acquiring fee, a weekly advance repayment and a monthly software subscription all repeat, and only the counterparty and the size separate them. Read the groups rather than trusting a count, and if a group is unfamiliar, look up the counterparty before you decide what it is.

Getting the twelve files into one place

1. 1

   Inventory the periods

   List all twelve statement periods for every account the business operates, so a missing month is visible before you start.
2. 2

   Convert each statement

   Convert every statement document, keeping one converted file per source statement so each row can be traced back to its page.
3. 3

   Chain the balances

   Confirm each file's closing balance equals the next file's opening balance and the periods run back to back, then resolve any break before combining.
4. 4

   Label the non-revenue credits

   Tag owner contributions, inter-account transfers, drawdowns and refunds so the deposit total reflects revenue only.
5. 5

   Compute the four measures

   Average daily balance per month, deposit count and adjusted total per month, negative days, and any fixed recurring debt service.

The chaining step is the one that protects everything downstream. Twelve files with twelve month names can still be missing a period, and the closing-to-opening chain is the strongest check available that no period between the first file and the last is missing or duplicated. It is the same step that anchors assembling a year of statements into one spreadsheet.

Where the business and its owner share an account, do the separation of business and personal transactions first. Personal spending inside a business account depresses every measure above, and a mixed account read by someone who has to guess at the split tends to get the conservative reading.

The measures are only as good as the extraction

Every figure in this article comes off the amount and balance columns. A single misread digit moves the running balance from that row to the end of the file, and an average daily balance computed from a corrupted balance column is a number that looks authoritative and is not.

The check is arithmetic, and it is the strongest automatic check available on a converted statement: recalculate opening balance plus credits minus debits row by row, and compare each result against the balance the statement printed. A misread digit cannot survive that, because the amount it corrupts feeds an equation whose answer the bank already printed alongside it, and the first row to disagree is the row that caused it. A dropped row cannot survive it either. Lose one of the 210.00 debits and the balance printed under it stops following from the row above, so the gap is flagged there and on every row after it until the file rejoins the printed column.

That is worth separating from the weaker check it gets confused with. Comparing a closing balance against the sum of the movements catches nothing that cancels: a dropped 210.00 debit and a duplicated one leave the closing figure exactly where it was, while the negative-day count and the daily average can both shift regardless, since neither is computed from the closing balance. Only the row-by-row comparison sees that pair, and only where the bank prints a running balance beside each transaction. Where it prints none, the totals comparison is all you have, and it deserves to be described as the weaker thing it is.

What does survive the row-by-row check is everything the equation never touches. Dates are not tested, and a date read wrong puts a balance on the wrong day, which is the one input average daily balance cannot be computed without. Descriptions are not tested, and the labelling of owner contributions and drawdowns above depends on nothing else. A zero-amount row moves no equation and is therefore invisible to it. So read a clean chain as settled evidence about the amount column and as no evidence at all about the rest of the row. For a lending file the amounts are what the decision is made on, and the labels are what you owe the reader.

Judged against that, hand-keying twelve statements gives you no control total of any kind: a transposed figure sits in the file silently, and the person who finds it is deciding how much of the rest to believe. The same limit applies to what the statement itself can carry, which is the movement of money, not the story behind it.