Blog · Finance metrics and formulas
AOV is net order revenue divided by eligible orders. Calculate it from an order export, compare channels on the same basis, and reconcile the result before deciding whether larger baskets improve profitability.
AOV meaning, in marketing, is average order value: net order revenue divided by eligible orders. It describes the dollars in an average basket after the adjustments your report includes. It does not tell you how many visitors bought or how much profit those orders earned.
Average order value = Net revenue from eligible orders / Distinct eligible orders
Net merchandise revenue = Gross merchandise sales - Discounts - Linked returns and allowances
Use the same orders in both parts of the calculation. For this article, an eligible order was completed during September 2026, belongs to a real customer, and was not canceled before completion. Its later linked merchandise refund reduces its revenue through the stated reporting cutoff. The original order remains in the count. A return record is an adjustment, not a new order.
This is a management reporting definition. The Census Bureau's retail definitions likewise describe sales net of refunds and allowances for returned merchandise and exclude sales taxes collected for governments. Here, shipping charges are excluded as an additional policy choice so every channel measures merchandise on the same basis.
Do not quietly change that definition when moving between reports. A checkout report including tax and shipping and a finance report net of merchandise refunds can both be useful, but their average values answer different questions. Label the revenue basis next to the figure.
Bring order ID, order date, customer ID, status, merchandise value, discount, refund amount, channel and currency. If the source has item lines, first sum those lines into one order. Count the distinct order IDs after eligibility filters; counting item lines measures a different denominator.
Keep a separate adjustment ID and original order ID for refunds. Check whether the merchandise amount is already discounted before subtracting the discount column. Otherwise a correctly recorded discount will reduce revenue twice.
The IRS instructions for Form 1120 distinguish gross receipts or sales from returns and allowances, including customer refunds and rebates. That supports keeping adjustments visible in the reconciliation rather than hiding them inside an unexplained sales total. It does not prescribe the operational AOV denominator.
Give every order one channel under the attribution rule used for this report. If the same order appears under several touches, either choose one owner or allocate both its revenue and its order weight consistently. The separate guide to marketing attribution models explains that choice; changing it can move channel AOV without changing the company's sales.
Consider a fictional US home-goods seller. The following table summarizes its eligible order rows for September 2026, with refunds recorded through October 15, 2026. All amounts are USD. These are constructed teaching inputs, not an industry benchmark.
| Channel | Gross merchandise sales (USD) | Discounts (USD) | Linked refunds (USD) | Net revenue (USD) | Distinct orders | AOV (USD) |
|---|---|---|---|---|---|---|
| 12,000 | 1,000 | 1,000 | 10,000 | 100 | 100.00 | |
| Paid search | 21,000 | 1,500 | 1,500 | 18,000 | 150 | 120.00 |
| Direct | 9,000 | 500 | 500 | 8,000 | 80 | 100.00 |
| Total | 42,000 | 3,000 | 3,000 | 36,000 | 330 | 109.09 |
Each net-revenue cell subtracts the two adjustment columns from gross sales. Paid search is 21,000 - 1,500 - 1,500 = $18,000; dividing by 150 eligible orders gives an AOV of $120.00. The business calculation is 36,000 / 330 = $109.09, rounded to cents.
For the layout above, put gross sales in column B, discounts in C, refunds in D, net revenue in E, orders in F and AOV in G. The row formulas are:
E2 = B2-C2-D2
G2 = IF(F2=0,"",E2/F2)
The total AOV uses the summed numerator and denominator:
=SUM(E2:E4)/SUM(F2:F4)
Leave a channel without eligible orders blank or label it unavailable. A displayed zero implies an observed average of zero, which is different from having no denominator.
First, the revenue bridge is 42,000 - 3,000 - 3,000 = $36,000. That must match the net merchandise revenue for the same order population and refund cutoff. Differences from the accounting period's ledger can be timing differences, so document them rather than forcing an artificial match.
Second, the channel counts sum to 100 + 150 + 80 = 330 orders. Compare that with the distinct eligible order list. A mismatch usually means an order has several channel labels, a blank label was dropped, or the item-line export was counted directly.
Third, weighting each channel's AOV by its order count reproduces the business result:
(100 × 100 + 150 × 120 + 80 × 100) / 330 = 109.09 USD per order
Use unrounded channel values in this check. The simple average, (100 + 120 + 100) / 3 = $106.67, is wrong for the business because it gives every channel equal weight despite different order counts. All displayed averages here round only after division.
Paid search has the largest basket in the example. That is a question to investigate, not a budget recommendation. Product mix, new versus returning customers, minimum-order promotions and refunds can all explain the difference. Compare the same segments before assigning the change to marketing.
Suppose paid-search refunds increase by $1,500 while every other input stays fixed. Its net revenue becomes 18,000 - 1,500 = $16,500 and its AOV becomes 16,500 / 150 = $110.00. Company net revenue becomes $34,500 and company AOV becomes 34,500 / 330 = $104.55. The order count did not change; a revenue adjustment moved the ratio.
Higher basket value also needs a margin test. Discounting or expensive fulfillment can make a larger basket less attractive. Use marketing ROI versus ROAS for the spending and margin comparison. Keep the distinct question of average order value and conversion rate separate: basket size and purchase completion need their own denominators.
Bring an order export with IDs, channel labels, merchandise amounts, discounts and linked refunds. Covirage maps the columns, and its tools compute and check net revenue, distinct order counts and AOV; the external AI model explains the result with every figure traced to its rows. Create a free Covirage account and bring a sample of your own export.
See customer profitability for the supporting account-level context.
State the revenue boundary before calculating. The example excludes sales tax and shipping charges to focus on merchandise. A business can report an alternative including shipping revenue, but it must label that version and use the same treatment across channels and periods.
No. AOV divides revenue by orders, while revenue per customer divides it by distinct customers. A customer can place several orders in a period. Keep both counts available so purchase frequency does not get mistaken for a larger basket.
Compare orders with similar products, customer types, refund treatment and dates. A higher AOV can come from a different product mix or a discount threshold. Check contribution after fulfillment and acquisition costs before treating the higher figure as an improvement.
Under the placed-and-completed-order policy used here, an originally eligible order remains counted and its linked refund reduces revenue. A fulfilled-order-only policy can differ. Document the rule, apply it consistently, and never count the refund itself as another order.