What is average order value (AOV)?

The metric describes buying behaviour per order, not per customer and not per visit. Someone who makes four small purchases in a month contributes four low-value orders, even if their commercial value is high. That is why AOV replaces neither customer lifetime value nor revenue per user, which answer different questions.

The delicate part is the word revenue. Google Analytics 4 keeps the concepts apart: item revenue excludes taxes and shipping costs, while total revenue in the monetisation report already subtracts refunds. Two legitimate figures that produce two different AOVs from exactly the same orders.

There is no single definition of an order either. Some shops count every confirmed transaction, others exclude the ones cancelled before dispatch, and those selling subscriptions have to decide whether each renewal is a new order. As long as the definition stays stable inside the same shop, the time series is usable. The moment it is compared against a competitor or an industry report, the comparison turns fragile, because what went into the numerator is almost never published.

It is worth remembering that this is an arithmetic mean and therefore sensitive to outliers. A handful of wholesale orders lifts the monthly AOV without anything having changed in the behaviour of the usual buyer. The median order amount, calculated alongside, exposes that effect immediately.

The base formula fits on one line.

AOV = ingresos / pedidos

All the work sits in deciding what belongs in each term. The version that compares best across periods starts from net product revenue and from the orders that were actually fulfilled.

ingresos netos =
    ingresos brutos
  - impuestos
  - gastos de envío
  - descuentos aplicados
  - importe de devoluciones

AOV neto = ingresos netos / pedidos

The gross variant, which adds taxes and shipping, is the one the shop dashboard usually produces by default and it serves treasury purposes. The net variant is used to decide marketing investment, because it comes close to the money available before the cost of goods. To go all the way to margin, there is a third version.

margen medio por pedido =
    AOV neto x margen bruto en %

A worked example, with invented figures to illustrate the method. A shop bills 62,400 euros in a month across 780 orders, of which 24 were returned for a total of 1,900 euros. Gross AOV is 62,400 / 780 = 80.00 euros. Deducting returns and excluding those orders leaves 60,500 / 756 = 80.03 euros. The figure barely moves here, but in categories with a high return rate the gap between both readings reaches double digits in percentage terms.

One timing detail remains. Revenue and orders must come from the same period and the same system. Mixing ERP revenue with orders counted in web analytics produces an AOV that matches neither source.

AOV decides how much a shop can spend to attract an order. Together with gross margin and acquisition cost it marks the point at which a campaign stops being profitable, and that point is calculated before launching it.

The second typical decision is the free shipping threshold. Setting it below the usual amount gives away margin on orders that would have happened anyway; placing it far above means almost nobody reaches it. Sensible practice puts it slightly above the most frequent order amount and then checks whether the distribution of orders has shifted towards that threshold.

It also organises conversion work. When traffic grows but revenue does not, AOV shows whether the problem lies in the number of orders or in their size. If orders rise and AOV falls, the campaign is attracting small-basket buyers, which is acceptable when those buyers come back and troublesome when they do not. That check requires looking at customer lifetime value as well as the isolated order.

• Write down what goes into revenue and what counts as an order, and keep that definition next to the dashboard where the figure appears.
• Calculate the median order amount alongside AOV. When the two figures diverge widely, extreme orders are moving the mean.
• Segment by channel, by device and by new versus returning customer before drawing conclusions. Overall AOV usually hides large differences.
• Deduct returns with the real time lag of your category. If the return window is thirty days, the current month's AOV is not final yet.
• Review AOV together with the number of orders. Lifting the mean by driving small buyers away can reduce total revenue.
• Recalculate the historical series when you change the definition, so that a change of method is not read as a change in the business.

• Comparing your own AOV against an industry figure whose calculation method is not published. Without knowing whether taxes, shipping or returns are included, the comparison says nothing.
• Mixing currencies in international shops without converting to a single currency at the exchange rate of the order date.
• Counting as orders the duplicate transactions created by a reloaded thank-you page. They inflate the denominator and sink the mean.
• Treating AOV as a profitability measure. A large order of low-margin products leaves less profit than a small one with a high margin.
• Setting an AOV target without setting an order-count target at the same time, which rewards lifting the mean at the expense of volume.