Ecommerce analytics: sales, customers and contribution

Ecommerce analytics compares online orders, returns, customer purchasing and contribution by product and channel. It helps a commercial team distinguish sales growth that retains value from sales growth consumed by refunds, product costs, fulfillment and promotion.

An order export is the starting point. Add the refunds and costs belonging to those orders before deciding which channel or product deserves more spending.

What ecommerce analytics covers

The US Census Bureau defines ecommerce around orders placed, or terms negotiated, through electronic systems; payment need not happen online. For a company review, make your own channel categories explicit so website orders, marketplace orders and other electronically ordered sales stay comparable.

The main questions are practical:

• Which products and channels generate net sales after discounts and refunds?
• How much of those sales remains after the agreed costs?
• Are buyers placing additional orders within the observation window?
• Which exceptions need a pricing, fulfillment or merchandising decision?

This is the broad commercial workflow. The neighboring ecommerce business intelligence guide covers the recurring reporting view; first establish the transaction definitions and checks that make that view reliable.

Define the files and the reporting window

Bring an order export with order identifier, fulfillment date, customer key, product, channel, gross item sales and discounts. Add a refund file that points back to the original order, a product cost file and reports of fulfillment costs, channel fees and advertising costs.

Use a stable internal customer identifier. Keep names and payment details out of the analytical file when the comparison only needs a customer key.

Choose a period basis. A calendar-month financial report includes refunds posted during the month, which may relate to older orders. An order-cohort review follows a defined set of orders and includes their subsequent refunds through a stated cutoff. Both are useful, but their totals can differ legitimately.

Net sales = Gross sales − Discounts − Refunds

Contribution = Net sales − Product cost − Fulfillment and channel costs − Assigned advertising cost

The treatment of refunds as a reduction of sales follows the net receipts principle described in IRS Publication 334. The contribution calculation is this article's management definition, not a claim that all companies classify their costs identically. The margin comparison guide explains why gross, contribution and net margins answer different questions.

Work one order cohort from revenue to contribution

A fictional US home-goods retailer reviews its July 2026 fulfilled orders, including related refunds posted through August 31, 2026. This is a small order-cohort example, not a calendar-month income statement. Each order contains one product, all amounts are USD, and sales tax is excluded.

Net sales are $700 − $30 − $170 = $500. The second table attaches each cost once to its order. E105 was returned in resalable condition, so its product cost has been fully reversed. Its fulfillment and advertising costs were still incurred.

Contribution is $500 − $220 − $100 − $90 = $90. Its margin is $90 / $500 = 18.0%. This amount remains before costs outside the stated scope, including fixed overhead.

Compare channels and products on the same basis

Summing the order rows produces the channel view:

Dollar averages are rounded to two decimals and percentages to one. Retain full precision for the calculations. Overall net sales per order are $500 / 6 = $83.33, using all fulfilled orders, including the order later fully refunded. The average order value guide covers the denominator in more detail.

The channels originally generated equal gross sales: $350 each, from their three order rows. After discounts, refunds and costs, their contributions differ sharply. The marketplace result calls for examining the returned bottle order and the associated costs before changing the entire channel's budget.

By product, mug sets contribute $76 on $350 of net sales, from orders E101, E102, E104 and E106. Bottle sets contribute $14 on $150, from E103 and E105. This is evidence to investigate returns and product experience. A tiny cohort cannot establish a stable product ranking.

Measure repeat purchasing without changing the question

There are four distinct buyers in the source table. C01 and C04 each have two fulfilled orders, so a repeat-buyer rate that includes subsequently refunded orders is 2 / 4 = 50.0%.

If the question is repeat purchasing with positive net sales, exclude fully refunded E105 before testing for a second order. Only C01 qualifies, giving 1 / 4 = 25.0%; all four buyers still have at least one positive-net order. Label which definition you use.

Neither figure is next-month retention. That requires an earlier customer cohort and a later observation period. The ecommerce cohort analysis guide owns that comparison.

Reconcile before assigning an action

The channel totals must tie: $320 + $180 = $500 of net sales and $88 + $2 = $90 of contribution. Product totals must reach the same figures: $350 + $150 = $500 and $76 + $14 = $90.

Tie the six order identifiers to the fulfillment register and the $170 refund total to the linked refund records. Reconcile advertising allocations to the source spending file, including any unallocated amount. To compare this cohort with financial reporting, show a bridge for refunds from other order periods and refunds posted after the reporting month.

Finance owns that bridge and the cost definitions. Ecommerce operations owns channel and order status consistency. Merchandising owns the product investigation, while marketing reviews assigned spending without assuming that the allocation proves causal impact.

Where it goes wrong

• Counting lines as orders. Multiple items or partial shipments can duplicate the order count and depress average order value.
• Missing later refunds. Comparing cohorts at different ages gives the newer cohort less time to accumulate returns.
• Losing return costs. Reversing revenue without retaining incurred fees and fulfillment costs overstates contribution.
• Mixing gross and net measures. Gross sales in one channel and net sales in another produce a misleading comparison.
• Forcing calendar and cohort totals to match. Their refund populations differ; reconcile the timing rather than overwriting it.

Compare your own ecommerce file in Covirage

Bring order and refund exports with matching identifiers, plus product costs and the channel costs you want included. Covirage maps the columns, and its tools compute and check the figures; the external AI model explains the results with every figure traced to its rows. Create a free Covirage account, which takes about a minute, then set up your business and bring a sample of your own data.

Use the result to support customer profitability decisions.