Seven reports cover everything a multi-channel seller needs to see monthly: settlement reconciliation, SKU level profit and loss, inventory valuation, customer acquisition cost by channel, a rolling cash forecast, sales tax liability against what marketplaces already collected, and a returns and reimbursement report. Run in that order, each one feeds the next. Skip the first and the other six inherit its errors.
Most sellers run two of these, usually a channel level P&L and something from an ad dashboard, and then wonder why the bank balance and the profit figure keep disagreeing.
1. Settlement reconciliation
This is the foundation report and the one most often skipped, because it is tedious and produces no insight on its own.
A marketplace deposit is a single net number covering thousands of transactions and a dozen fee types. The report has to decompose each settlement into gross sales, referral fees, fulfillment fees, storage, refunds, reimbursements, advertising deducted at source, reserves held, and tax collected. Every deposit that hit the bank should tie to one decomposed settlement, with nothing unexplained.
The timing makes this harder than it sounds. Amazon’s seller payments documentation states that it generally settles accounts every two weeks, that funds can take up to five business days to reach the bank after payment is initiated, and that it typically reserves funds for seven days on deliveries. So a settlement almost always straddles a period boundary, and the reserve is money earned but not received.
What good looks like: zero unreconciled deposits, and a reserve balance you can state from memory.
2. SKU level profit and loss
Channel level profit tells you Amazon made money. SKU level profit tells you which eleven products made it and which forty lost it.
The report needs, per SKU: units sold, gross revenue, landed cost of goods sold, referral and fulfillment fees, allocated storage, attributed advertising, and a returns provision. What falls out is contribution margin per unit and per SKU in total.
The number to watch is not the worst performer. It is the SKU with high volume and thin contribution, because that is the one absorbing working capital and warehouse space while returning almost nothing. Sellers usually know their obvious losers. They rarely know their volume traps.
What good looks like: a ranked list where the bottom quartile has a written decision next to it, reprice, rebundle, delist, or accept.
3. Inventory valuation and cost layers
Two figures matter here: what the inventory is worth, and whether any quantity is negative.
Negative quantities are not cosmetic. They mean the system recorded a sale of a unit it did not know it had, which forces it to guess the cost. Most systems guess with the last known cost or zero, and both answers corrupt gross margin for that period and every future one.
The valuation figure has tax consequences too. IRS Publication 538 requires that inventory practices stay consistent from year to year and that the method clearly reflect income, and a change in the method or basis used to value inventory generally requires filing Form 3115. A valuation report that swings around without explanation is a question waiting to be asked.
What good looks like: no negative quantities, and closing inventory that reconciles to a physical or cycle count within a tolerance you set in advance.
4. Customer acquisition cost by channel
Blended CAC, total marketing spend divided by total new customers, is a vanity number. It hides the channel that is working behind the channel that is not.
The report should split acquisition cost by channel and, where you can attribute it, by product. It also needs to be built from the books rather than from an ad platform, because ad platforms report their own attributed conversions and several platforms will each claim the same sale.
Building it from accounting data is a specific exercise, and the definitional choices matter more than the arithmetic: which costs count as acquisition, whether returning customers are excluded, and what period you match spend to revenue over. ConnectBooks has a walkthrough of doing exactly that from ledger data, which is the version your accountant will accept.
What good looks like: CAC by channel next to contribution margin per order, so you can see which channels clear their own cost.
5. A rolling cash forecast
Profitable inventory businesses run out of cash routinely, because inventory is paid for long before it is collected.
The forecast needs, week by week for the next 13 weeks: expected marketplace settlements with their real lag, supplier payments due, freight and duty, payroll, advertising, and any debt service. Model the lag honestly. Walmart Marketplace’s own seller guidance states that payment typically arrives about 28 days after an order ships. That is not an estimate you need to make, it is documented.
The output is one number: the lowest projected balance in the next 13 weeks, and the week it occurs. If that number is uncomfortable, you found out with eleven weeks to act.
What good looks like: a minimum balance you review weekly and a stated threshold that triggers a decision.
6. Sales tax liability versus marketplace collected
Marketplace facilitator rules shifted collection duty onto marketplaces for most seller transactions, which means much of the tax appearing in your data was never yours to remit. The report has to separate the two.
Split it three ways: tax the marketplace collected and remitted, tax you collected on your own storefront and owe, and tax neither party collected in a state where you may have an obligation. That third bucket is the one that becomes expensive.
Rules and nexus thresholds vary by state and change. The state’s department of revenue is the authority, and this is a question for a tax professional rather than a report you interpret alone.
What good looks like: a liability balance that actually gets drawn down by payments, rather than one that only ever climbs.
7. Returns and reimbursements
Returns are a variable cost, and treating them as a year end adjustment makes every monthly margin report optimistic.
The report needs return rate by SKU, the full cost of a return including the original fulfillment fee you do not get back, the share of returned units that were resold at full price versus liquidated, and separately, reimbursements owed to you for inventory the marketplace lost or damaged.
That last piece is money that goes uncollected constantly. It is also an area where a specialist tool beats a general one: Sellerboard audits Amazon for lost and damaged inventory and for cases where the reimbursement came in below the seller’s own cost, which is not something an accounting platform does.
What good looks like: return rate tracked per SKU as a trend, and a reimbursement claim balance that gets worked rather than admired.
Running the set
Order matters. Settlements first, because everything downstream reads from them. Then inventory, because cost layers feed SKU profit. Then SKU profit and CAC, which are the decision reports. Cash forecast next, since it needs the settlement timing. Tax and returns last, as they are reconciliation rather than steering.
Two hours a month is a realistic budget once the plumbing exists. Building the plumbing is the actual work, and it is mostly a data problem rather than an analysis problem: getting settlements decomposed into a ledger automatically instead of by hand.
If you only ever run one of these, run the settlement reconciliation. It is the least interesting report on the list and the only one that makes the other six trustworthy. The Small Business Administration’s guidance on managing business finances covers the general discipline, but the marketplace specific decomposition is on you.
