An e-commerce operator describes the business in gross merchandise value, because that is the number the dashboard shows and the number investors ask about. An underwriter has no use for it. The desk works from settled deposits: what the processor actually paid into the bank after refunds, chargebacks, fees and any reserve holdback, because that is the only money available to service debt.
The gap between those two figures is routinely fifteen to twenty percent, and occasionally much more. Operators who have never reconciled the two arrive in a credit conversation quoting a revenue figure the analyst cannot find anywhere in the statements, which starts the file on a bad footing. This article walks the reconciliation an analyst performs, what settlement timing and reserves do to a liquidity picture, how chargeback activity is read, and what to have ready so the review does not stall.
The reconciliation, top to bottom
Every e-commerce file gets walked down the same ladder. Know your own numbers at each rung before anyone else computes them.
None of those deductions is unusual or unfavorable. The problem is only ever the surprise. An operator who presents the ladder themselves, with the processor statements attached, is describing a well-understood business. An operator who quotes gross and lets the analyst find the difference has created a question about every other figure in the file.
Settlement timing is a liquidity fact, not an accounting detail
The lag between a customer's purchase and the funds arriving in your account determines what your bank statements look like, and bank statements are what get read. Two businesses with identical sales can present very differently purely on settlement mechanics.
| Channel | Typical settlement pattern | Effect on the bank statements |
|---|---|---|
| Card processing, established account | Daily batch, funds in one to two business days | Smooth daily deposit rhythm; easiest pattern to underwrite |
| Card processing, newer or higher-risk account | Delayed funding plus a rolling reserve | Deposits understate sales; reserve balance is invisible unless disclosed |
| Large marketplace payouts | Payouts on a fixed cycle, often biweekly | Lumpy deposits; average daily balance sags between payouts |
| Buy-now-pay-later providers | Settlement net of provider fees, varying schedule | A separate settlement stream that must be reconciled independently |
| Wholesale or retail partners | Net 30 to net 90 invoicing, sometimes with deductions | True receivables; the only channel where collection effort applies |
| Subscription billing | Recurring daily settlements, plus involuntary churn from failed cards | Predictable rhythm; failed-payment recovery affects real revenue |
The marketplace row deserves emphasis. A business paid every fourteen days runs its operating account down between payouts by design, which pulls the average daily balance well below what the annual revenue would suggest. That is a structural feature, not distress, and it is worth explaining in the file rather than leaving the analyst to interpret the sawtooth.
Reserves and holdbacks
Processors hold funds against future refund and chargeback exposure. A rolling reserve retains a percentage of each batch and releases it after a set period; a fixed reserve holds a stated sum outright. Either way, a portion of money you have earned is not money you can spend.
Disclose the arrangement and the current reserve balance. Undisclosed, it produces a permanent unexplained gap between reported revenue and deposits, which is one of the reconciliation failures analysts specifically look for. Disclosed, it is a normal feature of card acceptance, and the accumulated reserve balance can even be presented as an asset the business holds.
Chargebacks: the ratio that matters more than the dollars
Chargeback losses are usually small in absolute terms. Their significance is what they signal and what they threaten. A rising chargeback ratio puts the merchant account itself at risk, and a business that loses payment processing loses revenue entirely, not partially. That is why a desk reads the ratio as an existential exposure rather than a cost line.
The card networks operate monitoring programs with published thresholds, commonly discussed around the one percent level for chargebacks as a share of transactions, though the specific programs, measurement methods and thresholds differ by network and change over time. What matters operationally is the trend and the distance from the threshold, not a single month's figure.
What actually reduces the ratio
Most chargebacks in consumer e-commerce are not fraud. They are service disputes and unrecognized descriptors. A clear billing descriptor that matches the brand the customer bought from, delivery confirmation on every order, a refund policy that is easier to use than a dispute, and a responsive support channel resolve the bulk of them. Representment (contesting a chargeback with evidence) recovers some, but prevention is cheaper and it is what a reviewer wants to see documented.
Returns, and why the rate is underwritten
Return rates vary enormously by category, and a desk familiar with e-commerce knows the rough shape: consumables return at low single digits, apparel and footwear substantially higher. What draws attention is a rate that is high for the category, rising over time, or inconsistent with how revenue is recognized in the financial statements.
The accounting matters here. If returns are recorded as an expense rather than netted against revenue, the top line overstates real sales and margin looks better than it is. Analysts normalize this, and the normalization is easier and less suspicious when the operator has already done it. Present net revenue after returns as the primary figure and show gross separately.
Channel and platform concentration
Three dependencies get examined in e-commerce files, and all three are versions of the same question: what happens if the pipe closes?
- Sales channel. A business where a single marketplace produces the large majority of revenue is exposed to that marketplace's policy decisions, account suspensions and fee changes, none of which the operator controls.
- Payment processing. A single merchant account is a single point of failure for all revenue, as above.
- Customer acquisition. Heavy dependence on one advertising platform means acquisition cost is set by an auction you do not control. Rising acquisition cost against fixed gross margin compresses contribution quickly, and a desk that understands the model will ask about the trend.
None of these is disqualifying, and most successful e-commerce businesses have at least one of them. The productive response is to quantify each, state it plainly, and describe what is being done about it. Concentration you have named and are managing reads very differently from concentration the analyst discovered.
The document set
E-commerce files stall on missing platform data more than on anything in the financials. Most of it exports in a few minutes if you know to pull it.
What to have exported before the review
- Twelve months of processor statements showing gross volume, refunds, chargebacks, fees and net settlement by month.
- Marketplace or platform settlement reports for the same period, reconciled to the deposits in your bank statements.
- A monthly gross-to-net bridge: gross sales, returns, chargebacks, fees, net settled, in one table.
- Current reserve or holdback balance and the terms under which it releases.
- Chargeback count and ratio by month, with a note on what drove any spike.
- Return rate by month and by major category, with the accounting treatment stated.
- Revenue by channel, in dollars and percent, so concentration is visible without being asked for.
- Gross margin by channel after fees and shipping, since blended margin can hide a channel losing money.
- Inventory on hand at cost with days on hand, if the business holds stock rather than dropshipping.
- Six months of business bank statements for every account, so settlements can be traced end to end.
Which revenue figure should I put on an application?
Net revenue after returns, with gross shown separately if you want to present it. Using gross sets an expectation the bank statements will not support, and the correction happens in the least favorable way: during review, by someone who now doubts the rest of the file.
Will a lender contact my processor or platform directly?
Sometimes, and some programs read platform or processor data through a direct connection with your authorization. Requirements vary by lender and program. Assume the underlying data is verifiable and present it consistently rather than assuming any particular figure will be taken at face value.
My deposits look lumpy because of biweekly marketplace payouts. Does that hurt?
It affects average daily balance, which matters, but it is explainable and common. Include the payout calendar and reconcile each payout to the corresponding sales period. A sawtooth with a documented cause reads very differently from a sawtooth that has to be inferred.
How much do chargebacks need to matter before they become a problem?
The dollars are usually secondary; the ratio and the trend are what get read, because they bear on whether the merchant account survives. Network monitoring thresholds vary by program and change over time, so track your ratio monthly against your processor's stated expectations rather than a remembered figure.
I run mostly on one advertising platform. Is that disqualifying?
No, but expect it to be raised. Come prepared with acquisition cost trend, contribution margin after acquisition cost, and any second channel you are developing. The concern is not the dependence itself but whether the unit economics still work when acquisition cost rises, since that is the historical pattern.
Does dropshipping change how the business is underwritten?
Yes, in both directions. There is little inventory to fund and little inventory collateral to lend against, so the file leans harder on settlement history, margin and supplier reliability. Thin margins combined with no collateral narrow the structures available, and treatment varies by lender.
How much operating history is usually expected?
It varies considerably by lender and program, and shorter histories generally narrow the options rather than eliminating them. What consistently helps is a settlement record long enough to show a stable pattern through at least one seasonal cycle, since a single strong quarter is difficult to distinguish from a promotion that will not repeat.
Should I disclose a prior processor termination?
Yes. Processing history is verifiable, and a termination discovered later is far more damaging than one explained up front with the cause and the remediation. If the underlying issue was fixed: descriptor corrected, fulfillment delays resolved, refund policy changed, the documentation of that fix is the strongest part of the explanation.
Where to start
Build the monthly gross-to-net bridge for the last twelve months: gross sales, returns, chargebacks, fees, net settled. Then tie the net settled column to the deposits in your bank statements month by month. If the two columns do not agree, find out why before someone else does: the usual culprits are a reserve you forgot about, a second processor, or a channel settling into an account nobody included.
That single reconciliation is the most valuable page in an e-commerce credit file. It answers the question a desk was going to spend a week asking, and it lets the conversation move on to structure rather than arithmetic.
Qualified Commercial Underwriting Desk
Credit and capital markets
The Qualified Commercial underwriting desk reviews commercial real estate, dealer and Main Street files daily. The Academy is written from that work (how files are actually read, priced and declined) rather than from a rate sheet.
Educational content only. Nothing here is a commitment to lend, an offer of credit, or tax, legal or accounting advice. Program terms, timelines and thresholds vary by lender, file and market conditions, and any figures shown are illustrative.