A bookkeeper who does not understand ecommerce usually gives themselves away within one month end close. The signs are specific: they reconcile to the marketplace deposit instead of the settlement report, they cannot produce a cost of goods sold figure that moves with units sold, they treat every marketplace as one customer, they have no opinion about where sales tax sits, and they close the month without ever touching inventory. None of these mean the person is bad at bookkeeping. They mean the person is doing retail bookkeeping for a business that is not retail in the way their training assumed.
1. They reconcile to the deposit, not the settlement
This is the fastest tell. A general bookkeeper sees a deposit hit the bank, matches it to a revenue line, and considers the transaction reconciled. The bank agrees, so the work looks finished.
What that misses is everything the marketplace subtracted before sending the money. Amazon’s referral fees alone run from 5 percent to 45 percent by category, with most categories at 15 percent, according to Amazon’s published seller pricing as of the 2026 schedule, and referral is one line among storage, advertising, refunds, reimbursements, and reserve movements. A seller whose books reconcile to deposits has no fee expense recorded anywhere and no way to see which of those lines is growing.
Ask a prospective bookkeeper what they reconcile against. If the answer is the bank feed, the rest of this list will usually be true as well.
2. Cost of goods sold is a plug
In a services business, cost of goods sold is often a straightforward pass through. In a product business it is the number that determines whether the company makes money, and it has to move with units sold rather than with cash paid to suppliers.
The symptom is a COGS line that is suspiciously smooth, or one that spikes in the month a purchase order was paid. Both indicate that inventory is being expensed rather than capitalized. A bookkeeper doing this is not tracking the largest asset on the balance sheet. Gross margin by product does not exist, and it cannot be rebuilt later without going back through supplier invoices by hand.
The related tell is a bookkeeper who has no view on inventory valuation method. FIFO and weighted average produce materially different profit figures in a period when landed costs are moving, and choosing between them is a real decision, not a software default to be accepted quietly.
3. Every marketplace is one customer
A bookkeeper new to ecommerce will often set up Amazon as a single customer record and post one invoice a month. It reconciles. It is also useless.
Marketplaces behave less like customers and more like separate operating channels with their own fee structures, payout timing, return rates, and tax treatment. A seller running Amazon, Shopify, Walmart, TikTok Shop, and eBay has five channels whose economics diverge sharply, and a chart of accounts that cannot separate them produces a consolidated profit and loss statement that hides which channel is funding which.
This is the structural problem that ecommerce specific accounting tools exist to address. ConnectBooks, for instance, connects those marketplaces into QuickBooks Online, QuickBooks Desktop Enterprise, or Xero with channel and SKU level detail preserved rather than summarized away. Whether a seller uses software or a well built manual process, the requirement is the same: the detail has to survive the trip into the ledger, because it cannot be recovered afterward.
4. Sales tax has no home
Ask where collected sales tax sits. A bookkeeper who understands ecommerce will answer immediately that it is a liability, and will then ask which channels are covered by marketplace facilitator collection and which are not. A bookkeeper who does not will either say it is in revenue or will not be sure.
The second question matters more than the first. Marketplace facilitator laws shifted collection responsibility to the marketplace for most marketplace sales, but a seller’s own storefront is generally still theirs to handle, and nexus thresholds vary by state and change. That means a seller can be fully covered on one channel and fully exposed on another, simultaneously, without anything in the books signaling the difference.
Nobody should take a position on registration from a bookkeeper who has not raised the question, and the answer properly comes from a state department of revenue or a tax professional rather than from an accounting workflow.
5. The month closes without anyone counting anything
A close that never touches inventory is not a close. Units get lost, damaged, returned to sellable stock, disposed of by the fulfillment center, and occasionally reimbursed for at a value that has nothing to do with cost. None of that appears in a bank feed.
The sign is a balance sheet where the inventory number is identical to last month’s, or one that moves only when a purchase order is paid. Either pattern means the asset is being carried at a figure nobody has verified. Since inventory is usually the largest asset a product business owns, an unverified inventory balance makes the balance sheet as a whole unreliable, and a balance sheet nobody trusts is the single most common reason a seller’s diligence process stalls.
What to do about it
The fix is rarely a new bookkeeper. Most of the people doing this work are competent and have never been shown what is different about multichannel product businesses. The gap closes quickly when someone names it.
A reasonable first step is to hand over one month of raw settlement reports and ask for a reconciliation back to the deposits, with fees broken out by type. A bookkeeper who can do that has the instincts for the rest. One who cannot will say so, and that conversation is far cheaper in month three than in year three.
For sellers hiring rather than training, the AICPA maintains directories and credentialing standards worth checking against, and the Small Business Administration publishes a plain summary of what financial records a business is expected to keep. Neither is ecommerce specific, which is rather the point: the specialization is what a seller is looking for on top of the baseline.
