6 Signs Your Ecommerce Books Are Wrong

Your ecommerce books are wrong if any of these six things is true: the sum of your posted marketplace settlements does not equal the sum of your bank deposits, your gross margin percentage has not moved in a year, your 1099-K totals do not reconcile to your revenue accounts, inventory on the balance sheet has not changed since the last count, refunds show up as negative sales rather than their own line, or your accountant asked for a reserve balance and you did not know what that meant. Each one is a symptom with a specific cause. Here are the six, what each hides, and the fix.

1. Settlements posted do not equal deposits received

This is the first check a bookkeeper runs and the one most sellers have never done. Add up every settlement you posted for the month, per channel. Add up every deposit from that channel in the bank feed. They should match to the cent. When they do not, one of three things happened: a settlement was posted twice, a settlement was skipped, or a deposit was categorized as revenue directly without a settlement behind it.

Amazon makes this harder than it needs to be. Its settlement report documentation notes that settlement reports cannot be requested; Amazon schedules them. A report that arrived late or was archived is a settlement missing from your books until someone notices. Run the check monthly and the gap never grows past one period.

2. Gross margin percentage is flat

A margin that reads 41 percent in January and 41 percent in December is a margin someone typed in. Real cost of goods sold moves. Freight rates shift, suppliers reprice, a promotion pulls forward sales of your cheapest SKU, and each of those changes the blended margin. If yours does not change, COGS is being calculated as a fixed percentage of revenue, which means gross profit is fiction and every line below it is too.

The fix is landed cost per unit, layered FIFO so the September container’s price does not overwrite August’s, and COGS recognized as units ship. The IRS requires inventory to be valued at the start and end of the year under a consistent method, per Publication 538, and a flat percentage is not a method.

3. The 1099-K does not reconcile to revenue

Every marketplace and payment processor that handled your money sends a Form 1099-K reporting gross payments. The IRS explains on its 1099-K page that the form is required above $20,000 in more than 200 transactions and may arrive at lower amounts, and that you must report all income regardless. Gross on the form will be larger than the revenue in your books if your books are net of fees, and larger still than bank deposits. That is expected. What is not expected is a bridge that does not close: gross per the form, minus fees, minus refunds, minus reserves held, should land on your net revenue, and if it does not, one of those four lines is misstated.

Sellers who cannot produce that bridge in an afternoon have books that will not survive an examination. Build the reconciliation once, per channel, and rerun it every January.

4. Inventory on the balance sheet has not moved

Open the balance sheet and look at the inventory asset line for the last six months. If it reads the same number, or changes only when someone does a physical count, inventory is not flowing through the books. Purchases are hitting COGS when the bill is paid, which craters profit in the month a container lands and inflates it the month after, and neither month reflects what sold.

Inventory belongs on the balance sheet at cost when received, and it moves to COGS unit by unit as goods ship. A seller with multiple warehouses and an FBA balance needs the asset line to reflect all of it, including units in transit. A number that never moves is a number nobody is tracking.

5. Refunds are booked as negative sales

Netting refunds against revenue hides the refund rate, and the refund rate is one of the two or three numbers that decide whether a SKU is profitable. A product with a 38 percent margin and a 12 percent refund rate is a different business from the same product at 3 percent refunds, and a P&L that shows only net sales cannot tell them apart.

Refunds need their own contra-revenue line. So do the pieces around them: the marketplace’s refund administration fee where one applies, the returned unit going back into sellable stock at cost or being written off as unsellable, and any reimbursement the marketplace pays for lost or damaged goods. Amazon’s settlement carries each of those as a distinct amount type. Books that collapse them into one negative sales line have thrown away the detail.

6. Nobody can explain the reserve

Marketplaces and payment processors hold back part of what they owe you. Shopify’s payout documentation has a section on reserves, funds held temporarily to cover disputes and refunds. Amazon holds a portion of each settlement and releases it in a later one. That held balance is money you have earned and not received. It belongs on the balance sheet as a receivable.

If your books show the deposit as revenue and nothing for the reserve, revenue is understated in the settlement period and overstated when the release arrives. At low volume the error washes out. At high volume, with reserves in the tens of thousands, a month-end close can be off by a full week of sales. The sign that this is wrong is simple: ask what the current reserve balance is, and if the answer is a shrug, the books are missing an asset.

What all six have in common

Each one comes from posting what the bank shows instead of what the marketplace reports. The deposit is one number. The settlement behind it is hundreds of lines, and every one of the six signs above is a line that got collapsed into the total. Tools that read the settlement rather than the bank feed, and that carry unit-level cost alongside it, exist to prevent this. ConnectBooks is one, syncing Amazon, Shopify, Walmart, TikTok Shop, and eBay settlements into QuickBooks Online, QuickBooks Desktop Enterprise, or Xero with FIFO cost per unit. A2X and Bookkeep do the settlement half with COGS as a summary entry. Any of them beats a bank-feed-only ledger.

The test is not which software you run. It is whether you can pick one month, one channel, and tie the settlement to the deposit, the units shipped to the COGS posted, and the gross on the 1099-K to the net in your revenue account. If all three tie, the books are right. If any one does not, you now know which of the six to fix first.