Ad spend gets misclassified in ecommerce books in five recurring ways: it is netted out of revenue instead of recorded as an expense, it is lumped into one marketing account with no channel detail, it is split inconsistently between brand and performance spend, agency and tool fees are mixed in with media cost, and spend is recorded when the invoice arrives rather than when the advertising ran. Each one breaks a different decision, and together they explain why so many sellers cannot answer whether their advertising is working.
1. Netted out of revenue instead of expensed
Marketplace advertising is usually deducted from the payout rather than billed separately. A seller who books the deposit as revenue has absorbed the advertising cost into a smaller revenue number, and the books contain no marketing expense at all.
The effect is a business that appears to spend nothing on advertising while spending a great deal. Revenue is understated, marketing expense is zero, and any ratio built on either figure is meaningless. A seller in this position cannot calculate a return on ad spend from their own financial statements, which is why so many rely entirely on the marketplace’s advertising console and never reconcile it to anything.
The correction is the same one that fixes several other problems at once. Revenue posts at gross from the settlement report, and each deducted item, including advertising, posts to its own expense account, with the deposit as the reconciling total.
2. One marketing account for everything
Consolidating all advertising into a single marketing expense line is tidy and useless. A seller running Amazon sponsored products, Google, Meta, TikTok, and an email platform has five channels with different economics, different attribution windows, and different roles in the funnel.
Collapsed into one figure, the only question the books can answer is whether total marketing spend went up or down. They cannot show that one channel’s cost per acquisition doubled while another’s held, which is usually what is happening when blended performance deteriorates.
Channel level subaccounts cost nothing to create and cannot be reconstructed retroactively once a year of transactions has been posted to a single line. This is worth setting up before it is needed rather than after.
3. Brand and performance spend split inconsistently
Brand advertising and performance advertising do different jobs and should be evaluated on different timescales. Performance spend is judged on near term return. Brand spend is an investment whose payoff appears later and diffusely, if at all.
The misclassification is rarely deliberate. It happens when the same campaign serves both purposes, or when a seller reclassifies spend after the fact based on how it performed. Underperforming performance campaigns quietly become brand investment, which removes them from the return calculation and makes the remaining performance spend look better than it was.
The discipline is to classify at the point of spend, in writing, and leave it alone. A campaign that was performance spend in March is still performance spend in June regardless of what it returned.
4. Agency fees and tools mixed into media cost
Media spend is what reaches the platform. Agency retainers, management fees, creative production, and software subscriptions are costs of running the advertising function, not advertising itself.
Mixed together, return on ad spend is computed against a denominator that includes costs unrelated to media delivery, which understates the efficiency of the media and hides how much of the budget is being consumed by overhead. A seller paying a $4,000 monthly retainer on $20,000 of media has a materially different cost structure from one spending $24,000 on media directly, and blended into one line the two are indistinguishable.
Separate accounts for media, agency fees, creative, and tooling make both numbers visible: what the advertising cost, and what the advertising operation cost.
5. Recorded when invoiced rather than when the advertising ran
Advertising platforms bill on their own cycles, and marketplace advertising is deducted on settlement timing that does not align to calendar months. A seller recording spend when the charge lands is attributing cost to the period the money moved rather than the period the campaign ran.
In a flat month this is harmless. Around a promotional period it is not. Spend that drove November sales landing in December books makes November look efficient and December look expensive, and both conclusions are wrong. A seller reading those two months will draw the opposite lesson from the one the data supports.
The correction is an accrual: recognize advertising in the period it ran, and reverse it when the charge appears. The general rules on matching income and expense to the correct period are covered in IRS Publication 538, and the principle matters more here than in most expense categories because advertising is evaluated against revenue that is itself period sensitive.
What correct classification makes possible
The reason to fix these is not tidiness. It is that four specific questions become answerable.
What did it cost to acquire a customer on each channel, separated from what it cost to run the advertising operation. Which products are profitable after their own advertising rather than after a pro rata share of a blended total. Whether a given month’s performance reflects the campaigns that ran in it. And whether total marketing spend is growing faster than the revenue it produces, which is the question that decides whether a growth plan is working.
None of those can be answered from a single marketing line fed by net deposits.
Getting the inputs
The obstacle is usually data rather than intent. Marketplace advertising arrives buried in settlement reports, off platform channels bill separately, and attributing spend to specific products requires joining advertising data to sales data that lives somewhere else.
Tooling helps here. ConnectBooks, for instance, carries marketplace settlement detail into QuickBooks or Xero with SKU level profit and loss, which is the layer where advertising cost and product revenue can be compared without a manual join. Sellers also do this with a combination of a general ledger and a reporting layer, and plenty do it in a spreadsheet at smaller scale. The method matters less than the requirement: advertising spend has to be separable by channel and attributable to products, and neither property can be recovered once the data has been posted as a single monthly figure.
One practical caution applies to the ratios these classifications enable. Advertising efficiency measures are internal management figures rather than reporting standards, and they depend entirely on the classification choices described above. Two sellers quoting the same return on ad spend may be computing it differently enough that the numbers are not comparable, which is worth remembering before benchmarking against anything published by someone whose accounting is not visible.