Reading a Bank Statement Like an Underwriter

Most brokers read bank statements to confirm what the client told them. Underwriters read them to find out what the client did not.

That difference in purpose produces a difference in method, and the method is learnable. What follows is roughly the order an analyst moves through a file, and what each step is actually testing.

First: is this the business it says it is?

Before any number gets looked at, the header gets looked at. Account name, account type, address.

A file submitted for an LLC with statements in the owner's personal name is a different file. So is a business account whose address does not match the application, or a personal checking account being presented as business revenue. None of these are automatically fatal, but every one of them is a question, and a file that raises a question on page one gets read more suspiciously from then on.

Second: the shape of deposits, not the size

The average monthly deposit is the number everyone quotes. It is not the number that decides anything.

What gets read is the distribution. Twenty deposits a month of similar size reads as a business with many customers. Two deposits a month reads as a business with two customers, which is concentration risk regardless of the total. Deposits that arrive in a burst and then nothing for three weeks reads as project work, which changes how repayment gets modeled.

Then the trend. Three months roughly level is the easiest file to approve. Declining is the hardest, and a decline that nobody explained is harder than one that was explained. Rising sharply in the most recent month raises its own question — genuine growth, or money cycled in ahead of an application.

Third: the daily balance line

This is where an experienced analyst spends the most time, and where the least-prepared files fall apart.

The question is not how much came in, it is how much stayed. A business depositing $90,000 a month and holding an average daily balance of $800 is a business with no cushion, whatever the top-line says. It means every dollar that arrives is already committed, and a new fixed payment has nowhere to come from.

Negative days and NSF fees get counted here, and recency weighs more than volume. Six negative days spread across five months reads as ordinary small-business turbulence. Four negative days in the last thirty reads as a business that is tight right now, and "right now" is when the repayment would start.

Fourth: what is already going out

Existing positions from other funders are visible in the statements. Fixed daily or weekly debits, often on business days only, often in round or near-round amounts, usually from a recognizable set of originators.

They are always findable. The only variable is whether the broker disclosed them or the analyst discovered them.

A disclosed position is a math problem: can the business support another payment on top of this one. A discovered position is a character problem, and character problems end the conversation faster than math problems.

Stacking — layering new advances on top of existing ones until the daily burden exceeds what the business can carry — is also how a lot of small businesses get into real trouble. An operator who submits files that hide existing positions is not being clever, they are building a book that will default.

Fifth: the small things that change the read

• Transfers between the owner's own accounts counted as revenue. Analysts net these out; a file whose stated revenue assumed them will come back lower than expected.
• Payment processor deposits that arrive net of fees and chargebacks, so gross sales and banked revenue differ.
• Seasonality in a business where it is expected. A landscaper's February is not a warning sign. Not saying so in advance means the analyst has to work it out.
• A month missing from the submitted set. This is read as the worst possible month, every time. Send the consecutive months.

What this means before you submit

Read the statements the way the analyst will, and write down every question they raise. Then answer each one in the submission rather than waiting to be asked.

Three sentences explaining the June dip, the two existing positions with their balances and daily amounts, and what the money is actually for — that is usually the whole difference between an approval and a decline on an identical set of numbers.

And sometimes reading them properly tells you not to submit yet. A file with four negative days last month is a much better file after thirty clean days. Waiting is a legitimate strategy, and it is available only to the broker who read the statements before sending them rather than after the decline came back.