Ecommerce Analytics 101: Use Data to Sell Smarter

Every example and amount on this page is illustrative. It shows the method and is not a benchmark, a quote or advice.

What This Analysis Actually Measures

Profit contribution measures what an order, product or acquisition channel leaves behind after the costs that vary with selling it. The starting point is realized revenue: revenue from delivered orders, with cancelled and returned orders excluded rather than treated as successful sales. From there, the analysis accounts for product cost and other variable costs such as payment processing, fulfilment and shipping where those costs are available.

That result is more useful than sales revenue for deciding what to scale. A product can generate substantial revenue while contributing little after its cost of goods, discounts, transaction fees and fulfilment costs. A channel can also look efficient on attributed revenue while producing weak contribution once product costs and advertising spend are included. Fixed costs such as salaries, rent and software may still matter to total business profitability, but they should not be confused with the variable economics of a single sale.

The unit of analysis changes the question. Order-level contribution shows whether a particular basket was economically worthwhile. Product- and collection-level contribution shows which merchandise creates or destroys profit across many orders. Channel-level contribution connects the resulting sales economics with acquisition spend, while keeping measures such as MER and attributed ROAS distinct. This is the basis for a more practical ecommerce profit analytics view than a revenue-only dashboard.

ORVX uses a product cost model to derive gross profit, net profit and their margins, and reports profitability at product and collection level rather than only revenue. It also reports contribution after advertising by product and collection. The point is not to produce one universal score: it is to show the economic contribution at the level where the next decision will be made, in the merchant’s own trading currency.

Start With One Reliable Profit Metric

Use contribution margin as the operating metric for selling decisions, not revenue alone. Revenue tells you what customers paid; it does not tell you what remains after the costs created by those sales. Gross sales can look strong while discounts, refunds, product cost, payment fees, shipping, fulfilment or advertising leave little—or negative—contribution. A product that generates $10,000 in sales and $8,500 in variable costs contributes $1,500. That is the amount available to cover fixed costs and profit, not the $10,000 headline.

ROAS answers a narrower question: how much attributed revenue came from each advertising dollar. A campaign with $4,000 in attributed revenue from $1,000 of spend has a 4.0 ROAS, but that does not establish profitability. If the products sold carry $2,600 in product cost and other variable costs total $900, the campaign leaves $500 after advertising. ROAS can therefore support the analysis without replacing it. Keep MER—the store-level relationship between total revenue and total advertising spend—separate from attributed ROAS rather than treating either as a profit metric. Ecommerce Profit Analytics Software should make those distinctions visible.

Set the reporting period before comparing results: for example, delivered orders from 1–31 March and advertising spend assigned to that same period under one documented rule. Do not mix a month of orders with a different advertising window or count unfulfilled, cancelled or returned orders as realized revenue. Then calculate at the level of the decision: order for offer economics, product or collection for merchandising, and channel for acquisition. Review the total as a control, but do not let a profitable aggregate hide a loss-making product or channel. This consistent period-and-level setup is the foundation of Ecommerce Analytics 101, With the Numbers Made Clear.

The Formula, Stated Plainly

Use this formula for the amount an order, product or channel contributes before fixed operating expenses:

Contribution margin = realized revenue − variable cost of goods sold − other variable selling costs. Contribution-margin rate = contribution margin ÷ realized revenue × 100.

Realized revenue is the sales value retained from delivered orders: customer charges less discounts, refunds and other revenue reductions. Exclude cancelled and returned orders from the revenue base, and apply one consistent treatment to sales tax and separately charged shipping. Variable cost of goods sold is the product cost attached to the units sold, including the applicable unit purchase or manufacturing cost. Other variable selling costs are costs that rise when you sell or fulfil more orders: payment-processing fees, per-order fulfilment and packaging, shipping subsidies, marketplace commissions, return-processing costs and, when assessing acquisition economics, attributable advertising spend.

The cost boundary must match the decision. If you are comparing paid channels, include the ad spend assigned to each channel. If you are comparing product pricing before advertising, calculate a pre-ad contribution margin and label it clearly. Do not mix the two versions.

• Variable costs change with the number of units, orders or sales value. A 2.9% payment fee on a $100 realized order is $2.90; 100 such orders produce $290 of that cost.
• Fixed operating expenses do not move directly with the next order in the period: salaries, rent, software subscriptions, insurance, accounting and warehouse overhead are examples. Subtract them later when calculating operating or net profit, not when comparing the contribution from individual orders.
• Example of the rate: $60 realized revenue minus $24 product cost and $12 other variable costs leaves $24 contribution margin. The contribution-margin rate is $24 ÷ $60 = 40%.

A Worked Example From Order to Channel

Take one delivered order with $120 of merchandise, a $20 discount and $8 shipping charged to the customer. The customer initially pays $108. One $40 item is later returned, so realized revenue is $108 − $40 = $68. The $68 is the amount to carry into the order-level calculation, rather than the original checkout total. Next, attach the costs that belong to this order: the product cost for the items kept, any outbound or return shipping cost the merchant absorbs, the payment-processing fee, and the advertising cost assigned to the order. If those costs are represented by C, S, P and A, the order’s contribution after advertising is $68 − C − S − P − A. The $8 shipping charge is already included in the $108 collected; it is not automatically profit, because the merchant may have paid a separate amount to ship the parcel. Likewise, the $40 returned item should not remain in realized merchandise revenue, even if the original order record still shows it. If the returned unit is resold, its later sale belongs to a new delivered order; if it becomes dead stock, that is an inventory consequence rather than revenue from this order. Now group the same calculation by acquisition channel. Add realized revenue and each applicable cost across delivered orders from search, social, email or other channels, then divide the channel contribution by its realized revenue to obtain the contribution margin. Keep total marketing efficiency ratio (MER), which compares total ad spend with total store revenue, separate from attributed return on ad spend (ROAS), which assigns revenue to an advertising source. Neither ratio replaces the order calculation: both can look healthy while returns, fulfilment costs or low-margin products weaken contribution. A product and collection view reveals where the shortfall originates, while a customer view can show whether this buyer’s later repeat behaviour changes the relationship’s value. That is the practical purpose of ecommerce profit analytics: moving from a checkout total to a traceable number, then checking what it does—and does not—explain.

The Data Mistakes That Distort the Number

Missing variable costs make an attractive order look profitable. If an order produces $100 after discounts but costs $42 in goods, $8 in payment fees, $10 in fulfilment and $15 in shipping, its contribution is $25 before advertising. Leave out fulfilment and shipping, and the reported contribution becomes $50. That $25 difference can change a scale decision. The same problem appears when product costs are outdated: a margin calculation using a $20 unit cost is wrong if the current replacement cost is $27.

Revenue definitions also need to stay consistent. Gross merchandise sales, sales after discounts, cash collected, and revenue after refunds are different figures. A store with $100,000 in merchandise sales, $10,000 in discounts and $12,000 in refunds does not have $100,000 of realized revenue. Comparing that gross figure with net costs understates the cost of making each dollar. For a clearer treatment of the underlying arithmetic, see Ecommerce Analytics 101, With the Numbers Made Clear.

Refund timing can distort both monthly results and product decisions. Suppose a $200 order is recorded as January revenue, then refunded in February. January appears $200 stronger and February $200 weaker unless reporting recognizes the refund against the original sale or clearly labels the timing difference. Returned units can also remain in product sales while the refund is counted elsewhere, producing a false margin. ORVX calculates realized revenue from delivered orders and reports return rate by units and refund value against revenue.

Attribution gaps create a different kind of error. If the store spends $10,000 on advertising and produces $40,000 of store revenue, MER is 4.0. If ad platforms claim only $25,000 of that revenue, attributed ROAS is 2.5. Neither number replaces the other: MER includes the store's total revenue, while attributed ROAS depends on the platform's tracking and attribution rules. Missing channels, direct traffic, privacy restrictions or untracked campaigns can make attributed revenue incomplete. ORVX tracks MER and attributed ROAS separately rather than presenting them as interchangeable.

Duplicate orders inflate both revenue and cost efficiency. A Shopify order imported twice turns $500 of sales into $1,000 while ad spend remains unchanged; reported ROAS doubles even though the business did not earn more. The same duplication can overstate units sold and make stock appear lower than it is. A reliable dataset must treat the order identifier as a transaction key, not simply add every row in an export.

Blended averages hide where the economics actually break. A store can show a 55% overall gross margin when one collection earns 70% and another earns 20%, or when a high-margin repeat customer offsets loss-making first orders. Product and collection profitability, customer-level profitability and contribution after advertising expose those differences. That is why ecommerce profit analytics should be read at the product, collection and customer level, not only from a store-wide average.

What to Do With the Answer

Turn the result into a rule before changing anything. Set a minimum contribution after advertising for each channel, product group or order type. For illustration only, if the floor is $20 and a product leaves $32 before ads but costs $14 to acquire, its $18 contribution after ads is below the floor. Do not scale it because revenue is growing; either reduce acquisition cost, improve the offer or fix the product economics.

Use the same logic for budget allocation. Increase spend only when contribution after advertising is above the floor and the result is based on delivered orders, not a campaign’s reported revenue alone. Reduce spend when the result is positive but below the floor. Stop or pause an activity when it remains negative after a defined review window and there is no specific economics fix to test. Compare MER with attributed ROAS: ROAS can support channel-level diagnosis, while MER checks total advertising spend against total store revenue.

For pricing and promotions, calculate the new unit economics before launching. A $60 product with $24 in variable costs contributes $36 at full price. A $10 discount reduces that to $26 before advertising, so the promotion needs enough additional contribution from extra orders to justify the $10 given up per order. The arithmetic shows the cost of the offer; it cannot predict how demand will respond. ORVX can report where a price experiment is worth running, but it cannot estimate price elasticity.

For product mix, prioritize items and collections by contribution, not sales rank. A high-revenue product with weak contribution may deserve a margin fix, while a smaller collection with stronger contribution may deserve more budget. Check that decision against inventory position and dead stock: discounting can release cash, but the markdown impact should be measured against the margin you surrender. Ecommerce profit analytics keeps these decisions tied to product, collection and advertising results.

• Scale: contribution after advertising meets or exceeds the operator’s floor, with delivered-order economics supporting the decision.
• Fix: contribution before advertising is positive but the product, promotion or acquisition cost pushes it below the floor.
• Stop or pause: contribution after advertising stays negative through the agreed review window and no defined test addresses the cause.

Ecommerce analytics becomes useful when it connects a reported sale to the money that sale actually leaves behind. A consistent contribution-margin view gives operators a defensible basis for changing prices, promotions, product mix and acquisition spend without mistaking growth for profitability.

General information only: tax, duty and import rules change and depend on your situation, so check with an accountant or the official source before acting.