Ecommerce accounting is the practice of keeping books for a business that sells physical products through online marketplaces and storefronts, where revenue arrives as batched payouts net of fees, inventory sits in warehouses the seller does not control, and the same transaction may be taxed by a marketplace in one state and by the seller in another. It is ordinary double entry accounting applied to a transaction pattern that standard bookkeeping training does not cover. The principles are unchanged. The mechanics are different enough that a competent general bookkeeper can produce confidently wrong books for years.
What makes it different from regular accounting
Four structural features separate it from bookkeeping for a service business or a traditional retailer.
Revenue arrives net, in batches. A marketplace does not pay per sale. It accumulates activity over a settlement period, subtracts its fees, holds a reserve, nets out refunds, and deposits the remainder. One deposit may represent thousands of transactions and a dozen distinct fee types. The deposit is not revenue, and treating it as revenue is the single most common error in the field.
Inventory is held elsewhere. Stock sits in marketplace fulfillment centers, third party warehouses, and shipping containers, often at the same time. The seller owns it and has to value it without being able to walk out and count it. Units also go missing, get damaged, and get reimbursed by parties reporting on their own schedules.
Cost of goods sold requires real work. In a service business, cost of goods sold is often a simple pass through. In a product business it determines whether the company makes money, and it depends on landed cost, valuation method, and accurate unit movement. None of those arrive automatically.
Sales tax is split. Marketplace facilitator laws shifted collection to the marketplace for most marketplace sales, while a seller’s own storefront generally remains the seller’s responsibility. The same business can be covered on one channel and exposed on another at the same time.
The vocabulary, briefly
Settlement report: the marketplace’s itemized record of a payout period, showing gross sales, each fee, refunds, reimbursements, and the net deposit. This is the source document, not the bank feed.
Landed cost: the full cost of getting a unit into sellable inventory, including the supplier price, freight, duty, and inbound fulfillment fees. A cost of goods sold figure built on the purchase order price alone understates cost, often by a meaningful margin on imported goods.
Gross versus net revenue: gross is what the customer paid; net is what the marketplace deposited. Books should carry gross revenue with fees as separate expenses, so that the 1099-K reconciles and fee trends are visible.
Marketplace facilitator: a marketplace legally responsible for collecting and remitting sales tax on transactions it facilitates.
Contribution margin: revenue less the costs that vary per unit sold. Distinct from gross margin, and generally the better basis for deciding which products to push.
A worked example of the core problem
A seller receives a $41,200 Amazon deposit, and the naive treatment books $41,200 of revenue. The settlement report tells a different story.
Gross sales for the period were $58,900. Referral fees took $8,835, which is 15 percent, the rate most categories carry according to Amazon’s published seller pricing as of the 2026 schedule, with the full range running from 5 percent to 45 percent by category. Fulfillment took $6,100. Storage and long term storage took $740. Advertising took $3,900, refunds reversed $2,400 of sales, and reimbursements added back $275.
Under the naive treatment, revenue is understated by $17,700, no fee expense exists anywhere in the books, the return rate is invisible, and advertising cost does not appear as marketing spend. Gross margin is unknowable because cost of goods sold has nothing meaningful to sit against. Every downstream decision inherits the error.
Under correct treatment, each of those lines posts to its own account and the $41,200 becomes the reconciling total at the bottom. The work is one time setup, repeated automatically thereafter.
What the accounting stack usually looks like
Most sellers end up with three layers. A general ledger, typically QuickBooks Online, QuickBooks Desktop Enterprise, or Xero, which holds the books of record. A connector layer that translates marketplace settlement data into ledger entries. And an inventory layer that tracks units and computes cost of goods sold.
The connector layer is where the products differ most. Some summarize each settlement into a single journal entry, which reconciles cleanly and is sufficient for many sellers. Others preserve SKU level detail into the ledger, which costs more in complexity and is what a seller needs if they intend to analyze profitability per product or survive a diligence process that traces reported figures back to units. ConnectBooks sits in the second category for sellers running Amazon, Shopify, Walmart, TikTok Shop, and eBay, with details at https://www.connectbooks.com/. A2X, Link My Books, and Synder occupy adjacent positions with different marketplace and platform coverage, and a seller on Sage or NetSuite, or selling on Etsy, should start with what supports their stack rather than with any single recommendation.
The decision that matters is not which brand. It is whether detail survives the trip into the ledger, because summarized data cannot be disaggregated later.
What the rules require
Accounting method is a tax question with real consequences. Whether a business may use cash basis, and how it must treat inventory, is governed by the Internal Revenue Code and explained in IRS Publication 538. There is a small business exception: for taxable years beginning in 2026, a taxpayer meets the gross receipts test under section 448(c) if average annual gross receipts for the prior three years do not exceed $32,000,000, a figure set in section 4.30 of Revenue Procedure 2025-32.
That exception describes what a return may report. It does not describe what a seller needs to run the business. A company that expenses inventory on its tax return still needs accrual style inventory tracking internally, or it has no idea which products make money. The two purposes are different and both are legitimate.
Sales tax obligations depend on where a business has nexus, which varies by state and changes. That question belongs with a state department of revenue or a tax professional, not with an accounting workflow or an article.
When a seller needs this
A single channel seller doing modest volume can operate for a while on careful spreadsheets. The point at which that stops working is usually the second marketplace, because two channels with different fee structures, payout timing, and tax treatment cannot be reconciled by hand at any reasonable cost.
The earlier signal is simpler. If a seller cannot say what a given SKU earned last month after every cost that touched it, the accounting is not yet doing its job, regardless of how tidy the bank reconciliation looks.
