How do you do bookkeeping for an Amazon FBA business?

Use accrual accounting, not just the bank feed

The single biggest decision is accrual versus cash basis. Cash-basis bookkeeping records revenue and expenses when money moves; accrual records them when they're earned and incurred. For an inventory business, accrual is the right answer because it matches the cost of a product to the period you sold it, and it stops your payout timing from distorting your numbers.

FBA punishes cash-basis books in a specific way. Amazon holds a reserve and settles on a roughly biweekly cycle, so a sale on the 30th of the month may not hit your bank until the 5th of the next month. On a cash basis, that revenue jumps to the wrong period. Worse, your inventory (often your single largest cash outlay) is purchased months before it sells, so a cash view shows giant losses when you buy stock and giant profits when you sell it, with no relationship to how the business is actually performing. Accrual smooths this by treating inventory as an asset until it sells, then moving it to cost of goods sold.

One honest caveat: if you're brand new, selling a trickle, and filing a simple tax return, cash-basis books plus a disciplined inventory count can carry you for a while. The switch to accrual gets harder the longer you wait, though, because you eventually have to restate inventory and unwind netted deposits. Most sellers who reach real volume wish they'd started accrual from the first purchase order.

If you're weighing amazon fba vs dropshipping, the bookkeeping forks right here. A dropshipper never owns stock, so there's no inventory asset on the balance sheet and COGS is simply what the supplier billed against each order. FBA sellers carry inventory for weeks or months before it sells, which is exactly why accrual matters more to them and why a dropshipper can often get away with much simpler books.

Record settlements as summarized journal entries

The core mechanic of amazon fba accounting is the settlement journal. Instead of importing thousands of individual orders, you take each Amazon settlement and book one summarized entry that breaks the payout into its components. The deposit that hits your bank is just the bottom line; the journal restores everything Amazon netted out along the way.

All of this assumes you sell through Seller Central. Amazon vendor central vs seller central is a genuine bookkeeping fork: vendors sell wholesale to Amazon and get purchase orders, invoices, and chargebacks, which behave like ordinary B2B receivables, while third-party sellers get settlements. If you run both, keep them in separate revenue accounts, because they don't reconcile the same way and blending them makes each one impossible to check.

Here is an illustrative settlement journal (numbers are made up to show the structure, not real Amazon rates). It's a plain double entry bookkeeping example: every debit has a matching credit, and if the entry doesn't balance to the penny, a line got dropped. Suppose a settlement deposits $7,200 to your bank. The journal that produces that deposit might look like this:

• •Debit: Bank / clearing account $7,200 (the actual deposit)
• •Debit: Referral fees $1,500 (expense)
• •Debit: FBA fulfillment fees $900 (expense)
• •Debit: Advertising $400 (expense)
• •Debit: Refunds $300 (contra-revenue)
• •Credit: Gross product sales $10,000 (revenue)
• •Credit: Sales tax collected $200 (liability, if Amazon passed it to you)

Build a chart of accounts that matches the settlement

Your chart of accounts is what makes the settlement journal possible. At minimum, create separate accounts for gross sales, referral fees, FBA fulfillment fees, storage fees, advertising, refunds and returns, reimbursements, and a sales-tax liability. Lumping all Amazon costs into one 'Amazon fees' line technically balances, but it tells you nothing, and the whole point of bookkeeping is to see where the money goes.

A clearing account is the unsung hero here. Because Amazon settles on its own schedule and the deposit date rarely matches the settlement period, you post the settlement journal to a clearing account, then match the actual bank deposit against that clearing balance. When the clearing account nets to zero, you know every payout has been accounted for. This is the same pattern whether you keep your books in QuickBooks or Xero; for the full setup including account mapping, see our guide to Amazon accounting in QuickBooks & Xero.

Don't forget the fixed costs that hide inside the same payout. The Professional selling plan subscription is deducted in the settlement like everything else, and it belongs in its own subscription or platform-fee account rather than buried with referral fees, because it doesn't scale with sales. Amazon publishes the current price of each selling plan on its own fee pages and changes it periodically, so pull the amazon seller monthly fee from there instead of trusting a figure in an old spreadsheet.

Keep the fee accounts aligned with how Amazon actually bills, not how a generic retail template thinks. Referral fees scale with revenue, fulfillment fees with units shipped, and storage with time and cubic feet. When each has its own account, a margin squeeze announces which lever moved. When they're lumped, all you know is "Amazon costs more now," which isn't actionable.

Track inventory and cost of goods sold separately

Sales bookkeeping is only half the picture. The other half is inventory: what you paid for the units you're selling, including the landed cost of getting them to an Amazon warehouse. On an accrual basis, you record purchased inventory as an asset on the balance sheet, and only move its cost into cost of goods sold (COGS) when the units actually sell. That's what produces a true gross margin.

Most FBA sellers underestimate COGS because they forget the costs beyond the unit price: inbound freight, customs duty, whatever a prep center for Amazon FBA charges per unit to label and polybag, and shipping into Amazon. Folding those into your per-unit landed cost is what separates a margin number you can trust from a guess. Since Amazon now reimburses lost or damaged FBA inventory based on your sourcing cost, that same landed-cost record does double duty; it's both your COGS input and your evidence when Amazon owes you money for a lost unit.

Keep the unit-level detail out of your general ledger. The ledger should carry summarized inventory and COGS balances, while a subledger (a spreadsheet at small scale, amazon bookkeeping software at larger scale) tracks quantity and cost per SKU. Pushing thousands of SKU-level entries into QuickBooks is how sellers end up with unusably slow files, and it buys you nothing the subledger doesn't already provide.

Reimbursements: the settlement lines most sellers misread

FBA settlements include reimbursement lines whenever Amazon pays you back for inventory it lost, damaged, or mishandled. Book these to their own reimbursements account, not to sales. They aren't new revenue; they're recovery of inventory value you already owned, and mixing them into sales inflates your revenue trend while hiding your shrinkage rate.

The deeper reason to track reimbursements separately is that Amazon's automatic reimbursements don't catch everything. Lost inbound units, destroyed returns, and warehouse discrepancies all have claim windows, and money you don't claim inside the window is gone. If your books show what Amazon should have reimbursed versus what it actually paid, the gap becomes visible and claimable. Running a periodic FBA reimbursement audit against your inventory records is how sellers surface those unclaimed dollars; on real accounts the recovered amounts are frequently worth more than the bookkeeping time it takes to find them.

This is also where clean per-unit cost data pays off twice. Your reimbursement claim is only as strong as the cost evidence behind it, and the same landed-cost records that drive accurate COGS are exactly what a claim needs.

A monthly close routine that keeps FBA books honest

FBA bookkeeping stays manageable when it runs on a rhythm. A workable monthly close looks like this: post every settlement journal for the month, match all Amazon deposits against the clearing account, confirm the clearing balance equals only in-transit settlements, update the inventory subledger for received purchase orders and sold units, book the month's COGS, and review reimbursements against known inventory losses.

Then read the result like an operator, not a bookkeeper. Gross margin by month, fees as a percentage of sales, refund rate, and storage fees trending against inventory levels. Those four lines catch most FBA problems (a fee change, a return spike, an overstock quietly accruing storage) months before the bank balance does. A clean close also does most of the work on taxes for Amazon sellers before filing season starts, since gross sales already tie to the 1099-K, COGS is already booked, and the sales tax Amazon remitted on your behalf sits in its own liability account instead of inflating revenue. Write the close steps down as a checklist, too. A close that lives in one person's head skips steps under deadline pressure, and a skipped month is exactly when discrepancies take root.

The manual version of this close takes a few hours a month at moderate volume, and the settlement-splitting is the tedious part. That's the piece worth automating first. Most tools do it through an amazon seller central integration, authenticating against Amazon's selling-partner API and pulling settlement reports as soon as they're published. A good amazon quickbooks integration turns each settlement into a balanced journal automatically, and the better tools keep the SKU-level inventory subledger alongside so COGS and gross sales tie out without manual entry. Whichever connector you evaluate, test it on a real settlement before committing: post one month both ways, by hand and through the tool, and confirm the journals match to the penny. Compare candidates on inventory and reimbursement handling specifically, because the settlement-to-ledger core is table stakes and that's where the products actually differ.