The first question most owners ask of their analytics is simple: why does Google Analytics show fewer orders than Shopify? Sometimes the cause is a broken tag. Often it is consent: some visitors say no to tracking, and the numbers change as a result. The second kind of gap is normal, and you should know how big it is before you act on a chart.
This guide is for owners and managers who read GA4 reports and want to trust them. It explains which ecommerce events to check, how Shopify's own customer events line up with them and what consent mode and modelling change. I read Google's help pages and Shopify's developer pages on 11 November 2026. This is general information, not legal advice. For consent rules in your countries, ask an adviser.
The events GA4 expects
Google's list of recommended events names 14 ecommerce events. Here they are in shopping order, with what each is for:
- view_item_list: a person views a list of items.
- select_item: a person selects an item from a list.
- view_item: a person views an item.
- add_to_cart: a person adds an item to the cart.
- remove_from_cart: a person removes an item.
- view_cart: a person views the cart.
- begin_checkout: a person starts checkout.
- add_shipping_info: a person enters shipping information.
- add_payment_info: a person enters payment information.
- purchase: a person completes a purchase.
- refund: a person receives a refund.
- add_to_wishlist, view_promotion and select_promotion complete the list.
Google's page says these events fill the Ecommerce purchases report. Its page on explorations adds that the checkout funnel exploration uses begin_checkout, add_payment_info, add_shipping_info and purchase, can be broken down by country, and can show where people leave. It notes that a big drop at one step may mean a real obstacle such as a forced login, or a technical problem such as slow pages. It also depends on the events being set up first.
How Shopify's customer events line up
Shopify records its own standard customer events, which apps and pixels can use. The developer page lists page_viewed, product_viewed, collection_viewed, product_added_to_cart, product_removed_from_cart, cart_viewed, checkout_started, checkout_contact_info_submitted, checkout_address_info_submitted, checkout_shipping_info_submitted, payment_info_submitted, checkout_completed and search_submitted. By name, the closest matches are these:
- collection_viewed with view_item_list, and product_viewed with view_item.
- product_added_to_cart with add_to_cart, and product_removed_from_cart with remove_from_cart.
- cart_viewed with view_cart, and checkout_started with begin_checkout.
- checkout_shipping_info_submitted with add_shipping_info, and payment_info_submitted with add_payment_info.
- checkout_completed with purchase.
This is my reading by name. Whichever tool you use to send events to GA4 does its own mapping, so check the result, not the label. I did not read the documentation of any particular app, so I cannot tell you which events it sends.
How consent changes the picture
Shopify's Customer Privacy API handles four areas of consent: preferences, analytics, marketing and sale of data. The page says consent applies to Shopify-managed surfaces, including pixels, audiences and checkout. In regions set up to require consent, non-essential purposes are not allowed until the visitor agrees. Elsewhere, they are allowed by default. Shopify's own cookie banner appears in the configured regions and passes the choice to the API. Banners from apps work differently, so ask the vendor. The page also says the Global Privacy Control signal is honoured automatically in configured regions.
Google's consent mode page names two signals, analytics_storage and ad_storage. When they are denied, tags do not store cookies, but they send cookieless pings with coarse information that cannot identify a person directly, and Google uses them for modelling. There are two ways to set it up. In basic mode, Google tags are blocked until the visitor consents, so nothing is collected without consent. In advanced mode, tags load first and send cookieless pings if consent is refused. The page says behavioural modelling in Google Analytics is available with the advanced set-up, not the basic one, and that when consent is denied through a TCF-based consent solution, the property cannot model the missing data.
What modelling does, and what it does not
Google's help page says behavioural modelling uses machine learning to estimate the behaviour of visitors who decline analytics cookies, based on similar visitors who accept. It fills gaps in users, sessions, new users and key event rate. It applies to those metrics and not to event counts such as page_view. For your reports to show modelled data, set the reporting identity to Blended.
It is not automatic. A property must meet conditions, including at least 1,000 events a day with analytics storage denied for at least seven days, and at least 1,000 daily users with it granted on at least seven of the previous 28 days. Meeting them does not guarantee eligibility. A small store may not reach these numbers, which means you may see unmodelled gaps. And modelled data is not supported in audiences, user explorer, cohort and lifetime explorations, segments with a sequence, retention reports, predictive metrics or data export such as BigQuery. The page also warns that modelled data can make user counts higher than observed-only reports.
A routine to reconcile GA4 and Shopify
- Choose a week. Count orders and revenue in Shopify, and purchases and revenue in GA4. Write down the gap as a percentage.
- Do the same for three more weeks. A stable gap suggests consent and similar effects. A gap that jumps suggests a tag or a theme change.
- Place a test order with consent granted, and another with it declined, in a region where your banner appears. See what each does in GA4.
- Walk through the funnel as a customer and check each event fires once, in order.
- Check that the currency and value on purchase match the order.
- Record what you find and tell whoever reads the reports how large the gap is.
Other things to check when numbers differ
Consent explains part of a gap, not all of it. When the difference is larger than you expect, ask these questions in turn:
- Does the purchase event fire once per order, or sometimes twice or not at all?
- Are test orders and staff orders included in one report and not in the other?
- Do the currency and the value in the event match the order, including shipping and tax?
- Are refunds and cancelled orders treated the same way in both places?
- Did a theme update, an app install or a new checkout setting happen around the date the gap changed?
- Do orders placed in a shop or through another sales channel appear in Shopify but never reach GA4?
An invented example
A store sees 400 orders in Shopify one month and 340 purchases in GA4, a gap of 15 percent. The week-by-week gap holds steady between 13 and 17 percent. A test order with consent declined does not appear in GA4, and one with consent granted does. The owner concludes that most of the gap is visitors who declined consent, and reports GA4 revenue with a note that it is about 15 percent below Shopify. If the gap had jumped to 40 percent in one week, she would have checked what changed in the theme and apps that week instead.
How to use the numbers
Treat GA4 as a guide to direction, not an accounting record. For money, use Shopify's reports. Use GA4 for questions such as which channels bring people who buy, where they leave checkout and how countries differ. If you change your consent banner, expect a step in the numbers and note the date.
Our guide to privacy and cookies on a Shopify store covers the consent side. ESTASOFT's data and reporting page describes the reporting work we do. To have your set-up checked, send a short brief on the contact page.
Sources
- Recommended events (Google Analytics Help): list of 14 ecommerce events, Ecommerce purchases report; page date not shown (checked 11 November 2026)
- Ecommerce exploration solutions (Google Analytics Help): checkout funnel exploration, events used, breakdown by country, drop-off causes; page date not shown (checked 11 November 2026)
- Consent mode on websites and mobile apps (Google Analytics Help): analytics_storage and ad_storage, cookieless pings, basic and advanced mode, modelling; page date not shown (checked 11 November 2026)
- Behavioural modelling for consent mode (Google Analytics Help): how modelling works, eligibility numbers, affected and unsupported reports; page date not shown (checked 11 November 2026)
- Standard events (Shopify developer documentation): list of Shopify customer events including the checkout events; page date not shown (checked 11 November 2026)
- Customer Privacy API (Shopify developer documentation): four consent areas, regions, Shopify's banner, Global Privacy Control, checkout coverage; page date not shown (checked 11 November 2026)