E-commerce tracking in Google Analytics 4 with Google Tag Manager

Enhanced Ecommerce in GA4: what changed

If you are looking for Enhanced Ecommerce in GA4, this is it under a different name. Three things changed from Universal Analytics:

An old Universal Analytics data layer does not work as it is. The event names and the structure of the product array are different, so the site code has to change, not only the tags.

Which e-commerce events to track

Google’s developer guide groups the recommended events by step (Measure ecommerce, checked October 2026):

Start with four: view_item, add_to_cart, begin_checkout, purchase. They give you the funnel and the revenue. Add the list, promotion and shipping events when someone has a question that needs them. Every extra event is code to maintain.

Step 1. Push the events to the data layer

This is developer work on the site. Each action pushes an object with an event name and an ecommerce object. Google’s example for a purchase, shortened to one item:

dataLayer.push({ ecommerce: null });
dataLayer.push({
  event: "purchase",
  ecommerce: {
    transaction_id: "T_12345",
    value: 72.05,
    tax: 3.60,
    shipping: 5.99,
    currency: "USD",
    coupon: "SUMMER_SALE",
    items: [
      {
        item_id: "SKU_12345",
        item_name: "Stan and Friends Tee",
        item_brand: "Google",
        item_category: "Apparel",
        item_variant: "green",
        price: 10.01,
        quantity: 3
      }
    ]
  }
});

Four rules that prevent most problems:

• Clear the object first. The line dataLayer.push({ ecommerce: null }) removes the previous ecommerce object. Without it, items from an earlier event can leak into the next one.
• transaction_id must be unique and never empty. GA4 uses it to deduplicate purchases. The details are in duplicate transactions in GA4.
• value is the sum of price times quantity of the items, as Google’s example notes, and currency is required with it.
• Use the same item_id everywhere, from the list view to the purchase, or product reports will split one product into several.

Step 2. Create the tag in Google Tag Manager

You need the Google tag already on the site. If it is not, start with how to install Google Tag Manager and GA4 tags in GTM.

One tag can serve all e-commerce events.

1. Create a tag of the type Google Analytics: GA4 Event and enter your measurement ID.
2. In Event Name, choose the built-in variable {{Event}}. The tag then sends whatever event name the data layer pushed.
3. Open More Settings, then Ecommerce, tick Send Ecommerce data, and choose Data Layer as the data source.

Google’s guide describes this option as an alternative to mapping each parameter by hand with data layer variables. Use the manual mapping only if your data layer differs from Google’s structure.

4. Add a trigger of the type Custom Event. Tick Use regex matching and list the events you implemented:

view_item|add_to_cart|begin_checkout|purchase

5. Save the tag. Do not publish yet.

Step 3. Test before you publish

1. Click Preview in GTM and open your site in the debug window.
2. View a product, add it to the cart, start the checkout and complete a test order.
3. In the GTM debug panel, check that the tag fired once for each action and that the items array is filled.
4. In GA4, open Admin, then DebugView, and check that the same events arrive with their parameters.

Then publish the container. The e-commerce reports under Reports, Monetization fill within a day or two. GA4 treats purchase as a key event by default; the other events can be marked as key events under Admin, Events.

Common problems and their causes

How (not set) arises is covered in what (not set) means in GA4.

Refunds

Send a refund event with the transaction_id of the original order. For a partial refund, include the refunded items and quantities. Most shops process refunds in a back office, not in the browser, so this event is usually sent from the server with the GA4 Measurement Protocol.

Be realistic about it. Refund tracking in GA4 is rarely complete, and GA4 is not the system of record for money.

Where GA4 e-commerce reports stop

The setup above gives a marketing team what it needs for day-to-day work: funnel steps, product views, revenue by channel. An analyst who has to reconcile it with the shop system meets the limits quickly:

• GA4 revenue is what the browser reported. Blocked scripts, declined consent and closed tabs remove purchases. Orders by phone or marketplace never appear.
• Reports can be sampled or thresholded, and distinct counts are estimates, as described in data sampling in GA4.
• Product cost, margin and stock are not in GA4. Revenue by product is there. Profit by product is not.
• Returns arrive late or not at all.
• Explorations keep event-level data for 14 months at most on a standard property.

None of this is a reason to skip the setup. It is the reason the GA4 number and the finance number differ, and why nobody should try to make them equal inside GA4.

Exact funnel and product numbers from BigQuery

GA4’s native BigQuery export contains every e-commerce event with its items array. The funnel is then a query, counted from all events. This one runs as written on Google’s public GA4 sample:

SELECT
  event_name,
  COUNT(*) AS events,
  COUNT(DISTINCT user_pseudo_id) AS users
FROM
  `bigquery-public-data.ga4_obfuscated_sample_ecommerce.events_*`
WHERE _TABLE_SUFFIX BETWEEN '20210101' AND '20210131'
  AND event_name IN (
    'view_item', 'add_to_cart', 'begin_checkout',
    'add_shipping_info', 'add_payment_info',
    'purchase')
GROUP BY event_name
ORDER BY users DESC

Product revenue comes from unnesting the items of the purchase events:

SELECT
  item.item_name,
  SUM(item.quantity) AS units,
  ROUND(SUM(item.item_revenue), 2)
    AS item_revenue
FROM
  `bigquery-public-data.ga4_obfuscated_sample_ecommerce.events_*`,
  UNNEST(items) AS item
WHERE _TABLE_SUFFIX BETWEEN '20210101' AND '20210131'
  AND event_name = 'purchase'
GROUP BY item.item_name
ORDER BY item_revenue DESC

The first row on the sample is Google Crewneck Sweatshirt Green, 78 units and 3,542 in revenue. Linking a property is covered in connecting GA4 to BigQuery, and the query patterns in how to query GA4 event data in BigQuery.

The larger gain is what you can join. In the warehouse the GA4 events sit next to the orders from the shop system, with their real totals, refunds and costs, and next to ad spend. GA4 tells you which session and source led to an order; the shop system tells you what the order was worth.

From tracking to a model the business can use

Those joins are where an e-commerce team’s recurring questions live: revenue and margin by channel, repeat purchase by first product, what a customer from each paid channel costs and brings back. Answering them once is a query. Answering them every week, the same way, for people who do not write SQL, is a different job.

OWOX Data Marts is built for that job. An analyst defines the entities, such as orders, purchases, customers, sessions and traffic sources, as Data Marts on the warehouse, with SQL they wrote and descriptions the business can read. Reports on them are delivered to Google Sheets or Data Studio (the former Looker Studio) on a schedule. The E-commerce Data Model is a ready structure of such Data Marts to start from.

The honest boundaries: OWOX Data Marts does not track anything on your site and has no GA4 connector. The GA4 events reach your warehouse through Google’s own export, orders can be loaded from Shopify, and ad spend from the ad platforms. The SQL that turns events into sessions and orders into revenue is the analyst’s. The product publishes it, runs it on a schedule and keeps one definition in use.

Marketing AnalyticsGoogle Analytics for mobile apps: a GA4 and Firebase guideOctober 14, 2026→GA4Google Analytics 4 checklist and best practicesNovember 28, 2023→GA4GA4 Measurement Protocol: what it is and how to use itOctober 3, 2023→

See all articles →