What does the bank activity really say?
See deposits, revenue behaviour, balances, negative days, NSFs and cash-flow movement across the statement period — without reconstructing it line by line.
Deposits and revenue
A deposit is not revenue, and the statement will not tell you which is which.
One hundred and fourteen deposits across three statement periods, separated by what they actually are. Choose a category and the line itself comes off the statement, descriptor and all.
- 114
- Deposits in the period
- 06
- Categories they fall into
- 03
- It will not classify
The descriptor is the evidence
Choose a category to lift one of its lines off the statement. Nothing here rests on trusting a classification: the reason a transfer from the owner's other account is not a sale is visible in the line itself, which is a better argument than any figure attached to the same conclusion.
Why no revenue total appears here
Whether a supplier refund reduces revenue or simply is not revenue, whether owner contributions count at all, whether the proceeds of another advance are netted or ignored — these are policy questions, and different lenders answer them differently for good reasons of their own.
So the engine separates the deposits, labels what it can read, and stops. The total belongs to whoever set the policy, and a platform that produced one anyway would be making an underwriting decision in the least visible place it could possibly make it.
What the engine does with a deposit
- Reads the descriptorAlways
- Keeps the payer nameWhen there is one
- Decides what counts as revenueNever
- Guesses an unlabelled lineNever
Categories are applied from what the statement says, not inferred from what would be convenient. Where the descriptor is missing the deposit stays unclassified and visible, rather than being quietly absorbed into whichever bucket is largest.
Gross deposits is the number in every submission email and the one that is wrong most often — not because anybody lied, but because a bank statement makes no distinction between money a business earned and money that merely arrived. Separating those two things is most of the work, and deciding what to do about it is not ours.
The shape of the period
Six returned items across three statements is a count. Where they landed is the fact.
The period drawn as ninety-two days rather than three totals, with every negative day and every returned item on the date it happened.
- 92
- Days in the period
- 09
- Days closing below zero
- 05
- Returns on a day already below zero
May31 days
12345678910111213141516171819202122232425262728293031June30 days
123456789101112131415161718192021222324252627282930July31 days
12345678910111213141516171819202122232425262728293031
Everything, on the days it happenedNinety-two days drawn as ninety-two days. May is clean, July has a wobble at the very end, and June has a week in the middle that does not look like the rest of the period at all — and nobody had to be told that, because it is simply visible.
- 01Everything, on the days it happenedThe period as it actually is, rather than as three monthly totals. Totals are what hide a bad fortnight inside an acceptable quarter.
- 02Days the account closed below zeroNine days below zero. Where they sit turns out to matter far more than how many there are, and a count on its own cannot carry that.
- 03Returns that landed on a negative dayThe reading worth having. It is not a new measurement; it is two existing ones held against each other, which is the kind of thing that only becomes possible once the period is drawn as days.
What is reported, and what is not
- The date of every eventKept
- The statement page behind itAttached
- A cluster called out as a clusterYes
- A risk rating for the periodNone
A cluster of negative days is reported as a cluster of negative days. What it means — a seasonal trough, one late customer, a processor holding a batch — is not in the statement, and Cevrynt does not supply an answer the document cannot support.
Six returned items across three statements is true and almost useless. Four of them inside one week, five of them on days the account was already overdrawn, and eighty-three days either side that look completely ordinary — that is the same period, described in a way somebody can actually act on.
Existing positions
One advance you can see renewed, and one you can only see paying.
Each position drawn as a rail across the period, one tick per debit on the date it cleared. The rhythm is the evidence — and one of these rails has no funding credit anywhere on it.
- 02
- Positions on the statements
- 01
- With a funding credit visible
- 01
- Seen only by its payments
- $9,500 in · renewal proceedsHolidayHoliday
- No funding credit on these statements
The rhythm is the evidenceChoose a position to read what the rail shows. One tick per debit, on the date it cleared. A tick every business day is one kind of agreement and a tick every Tuesday is another — and the rail with no funding credit on it is exactly why it arrives as a question rather than a fact.
- 01Rapid Advance Funding, renewed on May 21The easy case, and still worth drawing: a daily debit with a lender's name on it, and a renewal credit from the same lender in the middle of it. 64 payments in the window, with the two bank holidays left as gaps rather than counted as failures.
- 02A weekly debit with no funding creditThe hard case. The payments are unmistakable and the origin is invisible, which is precisely the gap a reviewer needs to close by asking — and precisely the gap a platform should never close by assuming.
What the engine does with a debit rhythm
- Places every debit on its dateAlways
- Separates holidays from missesYes
- Confirms a position it cannot see fundedNever
- Stacking calculation or holdback adviceNone
A rhythm with no funding credit is reported as a rhythm with no funding credit. What it is, what else is owed on it, and whether it changes anything are questions for the broker and the reviewer — Cevrynt is not a lender and prices nothing.
One position arrived with its own name on it. The other has been paying every Tuesday for nine weeks without ever appearing on these statements as money received — and the second is the one worth an underwriter's first phone call of the morning.
What an average hides
The average is true, and the account still spent nine days overdrawn.
The daily closing balance as a curve, with the average laid flat across it. The distance between that line and the dip in June needs no annotation.
- $31.2K
- Average daily balance
- 09
- Days closing below zero
- −$8.0K
- Lowest close in the period
Move across the chart, or focus it and use the arrow keys, to read any single day
- MayAverage$39.4KDays below zero0Lowest close$30.9K
- JuneAverage$18.1KDays below zero7Lowest close−$8.0K
- JulyAverage$35.7KDays below zero2Lowest close−$2.0K
Why the curve and not the number
Average daily balance is in half the thresholds a lender writes, and it is precisely the kind of figure that is accurate and misleading at the same time. The average here is correct to the cent. So is the week in June, and one number has no way of carrying both.
Monthly averages do not rescue it either: June averaged comfortably above zero and still closed overdrawn on seven of its days. The only form that keeps both facts is the day-by-day series, which is why the engine keeps it and puts the average on top of it rather than in place of it.
What travels with an average
- The daily series behind itKept
- Days below zero, by dateListed
- The lowest close and whenStated
- An average reported aloneNever
The average is still reported, because lenders' policies are written against it. It simply never travels without the series underneath it, so the reader can see what it averaged over — and whether a comfortable number is sitting on top of an uncomfortable week.
$31.2K is the true average of this account, and the account could not cover itself for seven days in a row in June. Both are facts. A report that could only hold one of them would have to choose, and the one it would choose is the one that fits in a table.
Cash-flow movement
The totals nearly balance. Three weeks inside them do not.
Every section so far reads one side of the statement. Here is the other column: each week split at the axis, money in above and money out below, with the net ruled across it.
- $253.8K
- Money in across the period
- $257.5K
- Money out across the period
- −$3.7K
- Net, start to finish
Money inMoney outNet for the weekLoan repayments inside it
Two columns, held against each other
Choose a week to read its figures in whole dollars. Money in rises above the line and money out falls below it; the black mark is the week's net, which is exactly that week's change in the balance curve above — nothing here is estimated separately from it.
Why both columns, and why weekly
Everything earlier on this page reads one side of the statement. Deposits, the balance they produce, the repayments drawing on it — none of that shows what the business spends, and cash flow is only ever the two columns against each other.
Weekly is the grain where it shows. Across the whole period the totals nearly balance; inside it, three weeks had spending run at between two and a half and nearly five times income. The first two are back to back at the turn of May into June, and they drain the account into the overdrawn week the calendar and the curve already show. The third, in mid-July, does the same on a smaller scale and leads into the last two negative days.
What the engine does with outflows
- Reads both columnsAlways
- Reconciles to the balanceTo the dollar
- Separates loan repaymentsYes
- Judges what the spending was forNever
A week where more went out than came in is reported as that week. Whether it was a stock purchase, a late customer or something worse is not in the statement, and Cevrynt does not supply a reason the document cannot support.
Across three months the account took in $253,800 and paid out a little more than that. Both totals are true and neither is the story — the story is three weeks where spending ran at several times income, and you can only find them by holding the columns against each other one week at a time.
Structure and continuity
Three statements, whether they join up, and what they point at.
Everything above assumes one unbroken record of one account. Each statement’s arithmetic, each seam between them, and the account the file keeps referring to without including.
- 03
- Statements in the analysis window
- 02
- Seams that join to the dollar
- 01
- Account referenced, not submitted
Period covered · May 1 to July 3192 of 92 days · no gaps
- $39,720 = $39,720Close = next open
- $45,240 = $45,240Close = next open
Checked before anything else countsChoose a statement to read what its reconciliation says. Each one closes because opening plus money in minus money out equals the closing figure, computed here rather than asserted — and each seam joins because one statement's close is exactly the next one's open.
Why continuity comes first and gets shown last
Six statement files arrived — February to July, with March never supplied and July exported twice, as the Document Intelligence page shows. Everything on this page reads the most recent three months, May to July, and assumes those three are one unbroken record of one account. That is the first thing an analyst checks by hand, usually with a calculator and the first and last page of each statement, and it is the last thing a summary ever puts in front of anybody.
So it is shown in full: the calendar they cover, each statement's arithmetic written out, and the two seams where a close has to become an open. When all of that holds, nothing is reported. When it does not, that becomes the first finding on the file rather than a footnote under the last one.
What the continuity check covers
- Each statement's arithmeticReconciled
- Close into the next openChecked
- Days missing from the periodCounted
- Accounts referenced, not submittedListed
A statement that fails to reconcile, or a seam that does not join, is reported as exactly that and routed to a person. What it means — a missing page, a different account, an altered document — is a question the Fraud Signals side is built to put in front of a reviewer, not one this page answers.
Three statements, ninety-two days, two seams that join to the dollar — and one account the file keeps pointing at without ever including. None of that is exciting, which is exactly why it has to be checked before anything more interesting on this page is allowed to count.
Founder-led
Bring three months that argued with each other.
A file where the deposits looked fine and the balance did not, or where the monthly average was comfortable and one week inside it was not. Walk us through how you read it and what you asked for next, and we will show you the same period the way this works.