How to set up GA4 so it reconciles with Shopify and your store's orders
Filed
Baris Asa, Synctag, tracking specialist, London · Last checked:
GA4 and Shopify or WooCommerce usually disagree for two reasons nobody wrote down: which store figure GA4 revenue should equal, and which order number joins the two. Built to Google’s rule, GA4 revenue is product sales after discounts, with no tax or shipping. That equals Shopify’s gross sales minus discounts. Six decisions, made before the tags are built, keep the two matched. After that, the monthly check is a count, a total and a list of named orders.
This chapter is the build. If the numbers already disagree, the diagnosis is filed elsewhere:
- GA4 revenue higher than Shopify, or lower: every cause, normal and fault.
- Purchases missing products or value, or funnel steps at zero: F-301 and which field is broken.
- One order valued differently in GA4, the ad platforms and the store: F-302 .
- Finding which orders differ: Shopify’s export, GA4’s Transactions report and BigQuery.
Two terms. The yardstick is the store figure GA4 is meant to equal. The key is the order reference that lines up one GA4 purchase with one store order. GA4 calls it transaction_id.
| Decision | Settled in | Who |
|---|---|---|
| 1. The yardstick | A written definition | You, with tracking |
| 2. Clock and currency | GA4 settings and the tag | Tracking |
| 3. The key | The site’s code | Developer |
| 4. Product detail | The site’s code | Developer |
| 5. Refunds and cancellations | A rule, then maybe a server event | You, with tracking |
| 6. What GA4 never sees | A written list | You |
“Tracking” means whoever looks after GA4 and Google Tag Manager (GTM): an agency, a freelancer or an in-house analyst. “The site’s code” means the data layer, the order details the page hands to the tags when an order completes. You don’t build any of it. You decide 1, 5 and 6.
Is your setup already there? Three checks, about ten minutes, no GTM access.
- The key. In GA4, Reports, Monetization, Transactions, pick yesterday and look at the Transaction ID column. If the values look like your order numbers (#1001, or the order ID in your export), the key is in place. If they don’t, or some read (not set), see Decision 3.
- Tax and shipping. Take one order from that list. If GA4’s revenue for it equals the order total, tax and shipping are inside GA4’s value. If it equals the products after discount, they aren’t. Either is workable once it is written down (Decision 1).
- The count. Last month’s total of purchases in the same report against Shopify’s online store orders, test orders out. GA4 a little below is normal: shoppers who decline cookies or block tags aren’t counted. GA4 above is a duplicate (F-101 ).
Decision 1: which store figure should GA4 revenue equal?
Write it down before anyone builds a tag. The usual argument about GA4 and the store is two people reading different figures.
Google sets the GA4 side. The purchase value is the sum of price times quantity for the products, and Google says not to include shipping or tax, which go in their own parameters. Where a discount applies, Google says to send the discounted unit price. Built to Google’s rule, GA4 revenue is product sales after discounts.
The store figure that matches it is gross sales minus discounts:
- Shopify: in Analytics, Reports, the sales reports list gross sales and discounts as separate columns. Use the Online Store sales channel only (Decision 6). In Shopify’s own terms this is net sales with sales reversals added back, because net sales is gross sales minus discounts minus sales reversals, and excludes shipping and taxes.
- WooCommerce: in Analytics, Revenue, gross sales minus coupons. WooCommerce defines net sales as gross sales minus returns minus coupons, without tax or shipping.
Prices that include VAT. UK shops selling to consumers show prices with VAT in them, and that shelf price is often what the site hands to the tags. Google’s value has no tax in it, and Shopify’s net sales excludes taxes, so the price GA4 receives should be the ex-VAT price. Check one test order’s GA4 value against its subtotal before VAT.
If your tag sends the order total instead, with tax and shipping in it, the yardstick is the store’s total sales. Either can work. What fails is a tag built one way and a report read the other: revenue off by tax and shipping on every order. That is filed as F-302 .
Decision 2: which clock and which currency?
One time zone. GA4 reports in the property’s reporting time zone, in Admin, Property settings, Property details. Make it the store’s. Google says a change only affects data going forward and can leave a flat spot or a spike, so set it once and note the date if it ever moves. An order at 23:30 in London lands on the next day in a property set to Central European Time, and the month’s edges stop lining up.
The order’s own currency. Send the amount the customer was charged, with the three-letter code it was charged in, from the same place on every order. GA4 converts it to the property currency, so a multi-currency store keeps a small exchange-rate drift that is expected: what to send and why totals drift.
Decision 3: which order reference is the key?
An identifier that appears in your order export, sent unchanged as transaction_id on every purchase. A Shopify orders export carries both the order name (#1001) and a numeric order ID, so either joins. On WooCommerce, the order ID. What your tag sends depends on the app or code that sends it, so check: place a test order, and the next day find it in GA4 under Reports, Monetization, Transactions, by Transaction ID. If the value there can’t be found in your export, the key won’t join.
With the key in place, the order export and GA4’s Transaction ID column can be lined up order by order: matched, in the store only, or in GA4 only. Without it, all you can compare is totals.
Three things break it:
- An identifier the export doesn’t carry, such as a checkout token.
- An empty value. Google says GA4 deduplicates all purchases sent with an empty transaction_id, so real orders collapse into one.
- A different value per sender. Two senders with two IDs both count. Who may send the purchase is the deduplication chapter.
Decision 4: what product detail does the match need?
What you ask for: the SKU on every product line of every purchase. Order totals tell you whether GA4 and the store agree. Product detail tells you where they don’t.
Each purchase carries a list of products, the items array. Google requires item_id or item_name on each one. Send both, with item_id set to the identifier in your product export, usually the SKU. Then GA4’s Item revenue, product sales excluding tax and shipping, can be set against the store’s sales by product. Use the same refund rule as Decision 5: Google’s Item revenue has refunds taken off only when item-level refunds are sent, so compare it with gross sales minus discounts by product if refunds stay out.
Build the steps before purchase in the same release, with Google’s names exactly. A step never sent reads zero in GA4’s Purchase journey (view_item, add_to_cart, begin_checkout) or Checkout journey (add_shipping_info, add_payment_info), where shoppers really did leave. That is the funnel half of F-301 , listed event by event in which parameters each report needs.
Decision 5: do refunds and cancellations go to GA4, or stay out of the match?
Your call: gross or net. Pick one and write it down. Half-done is the version that can’t be reconciled.
- Leave them out. Compare GA4’s Gross purchase revenue, which Google defines as purchase revenue before refunds, with the yardstick from Decision 1. Nothing to build, as long as everyone reads the gross column.
- Send them. A refund event carrying the original order’s transaction_id, value and currency, plus the products for a partial refund. Google’s Purchase revenue is purchases minus refunds given, so it then follows the store’s net figure. Refunds are made in the store’s admin, not on a page with your tags, so in practice the event comes from the server: the store’s refund notification passed to GA4 through the Measurement Protocol, Google’s route for sending events from a server. The refund lands near the day it was made, not on the day of the sale, as it does in Shopify and WooCommerce.
The same rule covers the other direction: orders GA4 recorded at checkout that the store later cancels, voids or never collects payment for. Either they go to GA4 as refunds, or they are set aside on both sides of the match. How refunds reach GA4, and what goes wrong has the build.
Decision 6: which orders does GA4 never see?
GA4 only hears about orders that pass through a page with your tags on it. List the order sources that don’t, and take them out of the store side:
- Point of sale in a physical shop. Filter the yardstick from Decision 1 to the online store sales channel rather than the store total.
- Draft orders and orders keyed in by staff.
- Subscription renewals charged without the customer visiting the site.
- Marketplace and wholesale channels with their own checkout.
- Test orders. Shopify says test orders aren’t in its sales reports but are in the orders export, so take them out of the export.
Only whoever owns revenue knows which channels the business runs. Add a line the day a new one goes live.
How does the monthly match run?
The weekly count in the deduplication chapter is the quick check that catches a duplicate early. The monthly log is the record, and it tracks money as well as orders. It takes the last full month, at least two days after it ended, because Google says GA4 processing can take 24 to 48 hours and reports may change in that time.
- Count. The total row of GA4’s Transactions report (Reports, Monetization; if it’s missing, an editor can add it from the Library) against store orders, with Decision 6’s sources removed.
- Total. GA4 revenue against the yardstick, using the column your refund rule from Decision 5 names.
- Name the orders, only when a ratio moves. Join the order export to GA4’s Transaction ID for the same dates. How to read the result is Stage 5 of the tracking audit.
- Write one line. Both ratios, and what changed on the site that month.
Exhibit A: a reconciliation log for a UK online shop (illustrative figures, not a client). Each cell is GA4 as a share of the store.
| Month | Orders | Revenue | Changes on record |
|---|---|---|---|
| Aug | 91% | 92% | None |
| Sep | 91% | 92% | None |
| Oct | 108% | 109% | Upsell app, 2 Oct |
August and September are the first clean months after the build, and they set this shop’s band: GA4 about nine points below the store, on orders and money alike. Record that band in the log; it is this shop’s, not a number to copy. October is above the store on both, and the log puts the jump on 2 October. More purchases than orders is a second sender until proved otherwise: F-101 . If a ratio moves and the join still doesn’t name the orders, that is when I’m brought in: GA4 reconciliation.
Part of every band is shoppers who decline analytics cookies. A correctly wired Consent Mode keeps that part to the decline rate and nothing more, and lets Google model conversions in its attribution and ads reports once the volume is there.
On a store with tens of thousands of orders, the join outgrows a spreadsheet. GA4’s BigQuery export, its raw event data in Google’s data warehouse, carries the transaction ID on each purchase and the same join runs as a query. Google says a 2 to 5% difference in total event counts between the export and the reports is normal, and that the export’s time zone has to match the property’s.
What doesn’t reconciliation fix?
A clean match proves GA4 records the orders it can see, once each, at the agreed value. It doesn’t:
- Make GA4 agree with Google Ads or Meta. Each counts only the sales it can tie to its own ads, on its own windows: three purchase totals and which gaps are normal.
- Rebuild history. GA4 can’t backfill events it never received. The match is clean from the date the build went live, and the store’s reports are the record before it.
Keeping the written definitions from Decisions 1, 5 and 6 where the whole team can read them is the reporting chapter.
Common questions
How do I set up GA4 so revenue matches Shopify?
Decide which Shopify figure GA4 should equal and build the tag to it. Google’s rule for the purchase value is products only, at the discounted price, without tax or shipping, which matches Shopify’s gross sales minus discounts. If your prices include VAT, the price sent to GA4 must be ex-VAT. Send an order reference from the export as transaction_id, use the store’s time zone, and take point-of-sale and other off-site orders out of the Shopify side.
Should the GA4 purchase value include tax and shipping?
Not by Google’s rule. Google says to set value to the sum of price times quantity for the products and not to include shipping or tax, which have their own parameters. If your tag sends the order total instead, compare GA4 with the store’s total sales. The tag and the report you compare it with must use the same definition.
Which order ID should GA4 use as transaction_id?
One that is in your order export, so the two can be joined order by order. A Shopify orders export has both the order name (#1001) and a numeric order ID; WooCommerce has the order ID. Check a test order in GA4’s Transactions report. A checkout token can’t be joined, and an empty value makes GA4 merge orders together.
Should I send refunds to GA4?
Either send them or leave them out on purpose. Leaving them out works if you compare GA4’s Gross purchase revenue with the store’s figure before refunds. Sending them makes GA4’s Purchase revenue fall with the store’s net figure, but in practice the refund event comes from the server through the Measurement Protocol, because refunds are made in the store’s admin.
How do I record the normal gap after the build?
Take the first two or three full months after go-live and write GA4 purchases and GA4 revenue as a share of the store’s orders and yardstick in the log. That steady share is your band. Later months are read against it, with the site changes written next to them. Why GA4 sits below or above the store is in the GA4 and Shopify note.
How often should GA4 be reconciled with the store?
A count of purchases against orders weekly, and a full monthly log on the last full month, at least two days after it ended. Run the count again after every change to the theme, checkout, apps or tags. The order-by-order join is only for months where the ratio moves.
Also filed as
- ga4 shopify reconciliation
- how to make ga4 revenue match shopify
- ga4 woocommerce reconciliation
- how to reconcile ga4 with shopify
- set up ga4 so revenue matches shopify
- which shopify figure should ga4 revenue match
- ga4 transaction id order number reconciliation
- should ga4 purchase value include tax and shipping
Faults explained here
- F-302 Conversion value inconsistent Every platform reports a different revenue figure for the same orders: one includes shipping, one doesn't, one has its own currency idea. ROAS depends on who you ask.
- F-301 Ecommerce events partial Funnel blind spots: you can see purchases but not where shoppers drop off, or purchases arrive with no products or no value. GA4 sits below the store platform.
Sources
- Google for Developers: GA4 recommended events, purchase and refund (purchase value is the sum of price times quantity for the items, don't include shipping or tax; if a discount applies, set price to the discounted unit price; shipping and tax are separate parameters; refund requires transaction_id, value and currency for revenue metrics, items recommended for item-level refund metrics; item_id or item_name required on each item; checked 16 Nov 2026)
- Analytics Help: GA4 dimensions and metrics (Purchase revenue: purchases minus any refunds given; Gross purchase revenue: before refunds; Item revenue: items only, excluding tax and shipping, minus refunds; Ecommerce purchases; Transaction ID; checked 16 Nov 2026)
- Analytics Help: Minimize duplicate key events with transaction IDs (purchases with the same transaction ID are deduplicated; all purchases sent with an empty transaction ID are deduplicated; checked 16 Nov 2026)
- Analytics Help: About ecommerce metrics (event-scoped metrics count events, item-scoped metrics count items; the Ecommerce purchases report uses item-scoped dimensions and metrics; checked 16 Nov 2026)
- Analytics Help: Purchase journey report (session_start, view_item, add_to_cart, begin_checkout, purchase) and Checkout journey report (begin_checkout, add_shipping_info, add_payment_info, purchase); checked 7 Nov 2026
- Analytics Help: Set up Analytics for a website and/or app (reporting time zone and currency chosen per property; changing the time zone only affects data going forward and can leave a flat spot or a spike; checked 16 Nov 2026)
- Google for Developers: Measurement Protocol events reference (refund event: transaction_id required, value and currency, items recommended; checked 16 Nov 2026)
- Google for Developers: Measurement Protocol, send events (events can be backdated up to 72 hours; older events are accepted with the timestamp overridden to 72 hours ago unless validation_behavior is ENFORCE_RECOMMENDATIONS, which rejects them; checked 16 Nov 2026)
- Analytics Help: Data freshness (processing can take 24 to 48 hours; reports may change in that time)
- Analytics Help: Compare Analytics reports and data exported to BigQuery (some differences are normal, 2-5% in total event count is expected; match the export time zone to the property's reporting time zone; checked 16 Nov 2026)
- Shopify Help Center: Analytics fields (gross sales before discounts and sales reversals; net sales is gross sales minus discounts minus sales reversals; total sales adds fees, duties, shipping and taxes to net sales; checked 16 Nov 2026)
- Shopify Help Center: Finances report (net sales equates to gross sales minus discounts minus sales reversals and does not include shipping charges or taxes; sales reversals are the value of goods returned; total sales adds taxes, shipping and fees; checked 16 Nov 2026)
- WooCommerce: Analytics definition of terms (gross sales excludes refunds, coupons, taxes and shipping; net sales is gross sales minus returns minus coupons; total sales adds taxes and shipping; refunds dated on the day the return occurred; checked 16 Nov 2026)
- Shopify Help Center: Export orders (Name is the order number as it displays in the admin; Id is the order ID; checked 16 Nov 2026)
- Shopify Help Center: Sales discrepancies (test orders aren't included in sales reports but are included in the orders export; checked 16 Nov 2026)
When GA4 and the store won't agree, I build the tracking side of the six decisions with your developer, match one month order by order and leave the log your team keeps. Baris Asa, Synctag, London. For ecommerce brands and the agencies behind them.
GA4 reconciliation specialist →