The gap between how a founder reads a business and how a finance person reads the same business comes down to one thing: founders watch levels, finance watches rates of change. Revenue is up. Fine. Is it up faster or slower than last quarter, and did the cost to produce it move in the same direction? That second question is where the useful information lives, and it is the one that almost never gets asked at the founder level until something has already broken.
Below are the specific measures a competent ecommerce finance person watches weekly and most founders check quarterly or never.
Contribution margin by SKU, not gross margin overall
Blended gross margin is a number that hides its own worst cases. A catalog averaging 42% can contain a top-ten product running at 11% and the average will not move enough to notice.
Contribution margin at the product level takes unit revenue, subtracts landed cost, marketplace referral fee, fulfillment fee, storage, returns, and directly attributable advertising. What is left is what that product contributes to fixed costs. It is the only number that answers whether you should make more of something.
The reason founders skip it is that it is genuinely hard to compute. Marketplace fees arrive netted inside settlements rather than as itemized charges, so getting them to the SKU requires either a tool built for it or a person spending a day a month in spreadsheets. Platforms like ConnectBooks exist specifically to produce SKU-level profit and loss with cost of goods sold attached, which is the reason the category has grown even though QuickBooks and Xero both technically track inventory.
Cash conversion cycle, measured in days
Take days inventory outstanding, add days sales outstanding, subtract days payables outstanding. The result is how many days your cash is trapped between paying a supplier and getting paid by a customer.
For an inventory-carrying ecommerce brand, this number is usually somewhere between 60 and 150 days, and it is the single best predictor of whether growth will kill you. A brand growing 80% a year with a 120-day cycle needs financing. There is no version of that business where it does not.
Founders instead watch the bank balance, which is the cash conversion cycle expressed as a lagging indicator with no explanatory power. By the time the balance tells you something, the container is already on the water.
Reserve balance as a share of monthly payout
Marketplaces hold back funds against potential refunds and claims. The balance moves with sales velocity, refund rate, and account standing, and it is not visible anywhere in your ledger unless someone put it there.
A finance person tracks the reserve as a percentage of what you expect to receive. When that percentage climbs, it means either the marketplace’s model of your risk changed or your refund rate did. Both are worth knowing before the payout comes in short.
Fee band exposure
This one is new enough that most founders have not internalized it.
Amazon’s 2026 US fulfillment fee structure, published in Seller Central, applies different rates across three price bands: under $10, $10 to $50, and above $50. Amazon states that standard-size products priced above $50 see fulfillment fees rise by $0.31 per unit on average versus the prior card, while standard-size items priced below $10 rise $0.05 per unit on average.
The consequence is that a product’s fulfillment cost now depends on its list price. Amazon’s FAQ specifies that items priced at exactly $10 and exactly $50 fall into the middle band.
A finance person keeps a list of every SKU sitting within a few dollars of those boundaries, because a routine promotion on one of them changes the unit economics rather than just the top line. Amazon also documents a 3.5% fuel and logistics-related surcharge on FBA fulfillment fees in the US and Canada beginning April 17, 2026, and a peak fee window running October 15, 2026 through January 14, 2027. Neither is reflected in the published base rate cards.
Inventory turns by category, with a stale tail
Founders track total inventory value, which conflates two very different situations: a lot of fast-moving stock, and a moderate amount of stock that will not move at all.
The measure that separates them is turns by category, paired with an aging report showing what has sat past 180 days. The aging tail is where the write-down is hiding, and it grows quietly because nobody wants to be the person who says the 900 units of last year’s variant are worth less than the ledger claims.
Amazon adds a direct financial signal here. Its 2026 fee documentation describes a low inventory level fee charged on shipped units when inventory for standard-size and bulky products falls below 28 days of supply relative to customer demand. So the cost of being over-stocked and the cost of being under-stocked are now both explicit line items.
Return rate by SKU, trended
Aggregate return rate is a vanity metric. It moves slowly and averages away the products causing the problem.
Return rate by SKU, trended over rolling twelve weeks, catches a supplier quality change roughly two months before customer reviews do. It also catches listing problems, since a spike concentrated in one variant usually means the product page is describing something the box does not contain.
Advertising cost as a share of contribution, not of revenue
Advertising cost of sale measured against revenue is the standard marketplace metric and it is misleading, because a 20% advertising cost of sale is comfortable on a 55% margin product and fatal on a 25% margin product.
Measuring ad spend against contribution margin instead tells you what fraction of the actual profit you are handing to the platform. Some products will turn out to be advertising businesses that happen to ship goods.
The one founders get right
In fairness: founders are usually better than finance people at knowing which products customers love and which ones they tolerate. That knowledge is real and it does not show up in any of the measures above.
The failure mode is not that founders watch the wrong things. It is that they watch qualitative signals with high resolution and quantitative signals with almost none, and then make inventory bets that require both.
How to close the gap without hiring a CFO
Pick three of the measures above and put them on one page reviewed every Monday. Contribution margin for the top twenty SKUs, cash conversion cycle, and the 180-day inventory tail is a reasonable starting set for most brands under $20M.
The page matters more than the tooling. A weekly ritual with imperfect numbers beats a perfect dashboard nobody opens. Once the ritual exists, the argument for better data makes itself, because you will get tired of the same three questions being unanswerable.
Amazon’s fee documentation, Shopify’s payout documentation, and the IRS guidance on accounting methods and inventory in Publication 538 are the three primary sources worth reading before you set the page up. None of them are exciting. All of them will change a number you currently believe.
