GA4's definition of AOV
The GA4 developer guide to measuring ecommerce defines a set of recommended events, ending in purchase, that share an items array and carry the order's details. On the purchase event, value is the sum of each item's price multiplied by its quantity; tax and shipping are sent in separate parameters; currency states the unit; and transaction_id identifies the order so that the same purchase is not recorded twice. Google Analytics derives purchase revenue and the count of purchases from those events, and average purchase revenue divides the first by the second. Every input is therefore set by what the site's data layer and tags pass, which is why two shops with identical orders can show different averages in GA4.
What counts as revenue
Revenue is whatever the purchase event carries in value. Following the developer guide, that is the item total, so tax and shipping are not part of it, and discounts count only as far as the item prices sent already reflect them. A site that puts the order total into value instead reports a higher average for the same orders, and a site that changes convention between templates, for example between a standard checkout and an express one, blends two definitions into one figure. Refunds lower revenue only when the site sends a refund event with the original transaction ID; most refunds are processed in a back office, so many properties never send them and GA4's revenue stays at the pre-refund level.
What counts as an order
The denominator is the number of purchase events, with the transaction ID marking each one as a distinct order. A purchase sent without an ID, or with a placeholder, has nothing to identify it, so a reloaded confirmation page or a returning visitor can add another purchase. Orders that never send an event at all, because the visitor refused consent, a blocker stopped the tag or the payment finished on another domain and never returned, are missing from both halves of the division.
Whose orders are included
GA4 includes every purchase it records, from any channel, which makes its average a property-wide figure rather than an advertising one. Segmenting by session source, campaign or device narrows it to the orders those sessions produced, under the property's own attribution and reporting rules.
Where to find AOV in GA4
In the standard reports, the Monetisation section holds Ecommerce purchases, with item revenue and purchase counts, and the monetisation overview shows average purchase revenue as a card beside total revenue. In explorations, the metrics Purchase revenue, Transactions and Average purchase revenue can be laid against any dimension, which is where the average is broken down by channel, device or landing page. To see what the average is built from, open DebugView while placing a test order and read the purchase event's parameters: value, tax, shipping, currency and transaction_id. The GA4 reports and dimensions guide explains where these reports sit and how their scopes differ, and the ecommerce tracking tutorial shows how the purchase event is built.
Discrepancies with the shop platform and Google Ads
| Source of difference | GA4 | Shop platform | Google Ads |
|---|---|---|---|
| Revenue basis | Purchase event value, item total by the guide's convention | Order totals, often including tax and delivery | Value passed by the conversion action |
| Orders included | Purchases whose event reached GA4 | Every order taken | Conversions attributed to Google Ads only |
| Identity of an order | Transaction ID on the event | The order record | The conversion action's own counting |
| Refunds | Only if a refund event is sent | Booked against the order | Not seen by the conversion tag |
The shop platform's average usually includes tax and shipping, which GA4 excludes when the guide is followed, so the shop's figure tends to be higher per order; it also counts orders GA4 never saw because of consent choices or blockers. Google Ads offers its own version through value per conversion, but it describes only orders attributed to Google Ads, valued by the conversion action and credited under its own model, so it cannot stand in for an average across all customers. The AOV levers guide uses the shop's own figure for that reason when judging merchandising changes.
Two averages for one tea shop
Damsonplume Tea Merchants, an invented loose-leaf tea shop, reads a month in GA4: 1,240 purchases and £44,640 of purchase revenue, an average of £36. The shop platform shows 1,410 orders and £56,400 including VAT and delivery, an average of £40. An audit of transaction IDs finds 38 GA4 purchases that repeat an order already counted, sent when customers reloaded the confirmation page. The team fixes the confirmation tag, records that GA4's value excludes VAT and delivery by design, and stops treating the £4 gap as a problem.
Interpreting GA4 AOV
Read GA4's average for its trend rather than its level, unless the value convention has been checked against the shop's. It is most useful segmented: by channel, device, landing page or campaign, it shows where order size moves, which the shop platform often cannot. Before reacting to a change, check for edits to the purchase tag, a new checkout template or a change in consent rates, any of which moves the average with no change in what customers buy. Reconcile monthly with the shop for orders, revenue and refunds, and keep a short written note of the value convention beside every report that quotes the figure.
