Every store owner eventually puts two numbers side by side: revenue in the WooCommerce or Magento admin, and purchase revenue in Google Analytics 4. They never match. Sometimes GA4 is 8 percent low, sometimes 30 percent low, and once in a while it is higher than the store, which should be impossible.
A gap by itself is normal. The two systems measure different things. Your store database is a ledger of orders and payments. GA4 is a record of what browsers managed to report. The problem is an unexplained gap, because you cannot tell whether a drop in GA4 revenue means fewer sales or a broken tag. This guide breaks the gap into its parts so each one can be measured and, where it matters, fixed.
Start from the ledger, not from GA4
Treat the order database as the reference and ask how much of it GA4 should be able to see. The answer is never 100 percent. The chart below shows how a month of store revenue typically shrinks on its way into GA4. The sizes are illustrative; your own proportions will differ, and finding them is the point of the exercise.
The seven causes, and how to measure each one
1. Order status
The admin order list includes orders that never became revenue. In WooCommerce, a Pending payment order has been received but no payment has been collected, and an On hold order is awaiting payment confirmation even though stock has already been reduced. Failed and Cancelled orders sit in the same table. If your GA4 purchase tag fires on the thank-you page, some of these orders sent a purchase event and some did not, depending on the payment method. Bank transfer and cash on delivery are the usual suspects.
Decide which statuses count as a sale for this comparison (usually Processing and Completed, plus Refunded at original value) and filter the store side to those before comparing anything.
2. Refunds
GA4 only reduces revenue when it receives a refund event, and that event has to carry the same transaction ID as the original purchase. As one Magento integration guide puts it, no match means no correction. Most stores never send refunds at all, so GA4 shows gross revenue while the store’s net figure falls. Compare gross to gross first, then handle refunds as a separate line.
3. Time zones and date boundaries
WooCommerce’s order table stores date_created_gmt in UTC (it is part of the High-Performance Order Storage schema), Magento stores created_at in UTC, and GA4 reports in the property’s own time zone. A store in Toronto with a GA4 property set to Pacific time will move three hours of orders across every day boundary. Over a month the effect is small. Over a single day, especially Black Friday, it can look like a tracking outage.
4. Consent
When a visitor declines analytics cookies, GA4 receives either nothing (basic consent mode) or cookieless pings (advanced consent mode). Google can model the missing behaviour, but only above a volume floor: the property needs at least 1,000 events per day with analytics_storage set to denied for at least 7 days, and at least 1,000 daily users who granted consent on 7 of the previous 28 days. Most small and mid-sized stores never reach that floor, so for them declined consent is simply missing revenue. Your consent platform reports its acceptance rate; that rate is a first estimate of this step.
5. Ad blockers and browser privacy
Many blockers stop the GA4 script outright. GWI survey data put ad blocker use at 29.5 percent of internet users worldwide in Q2 2025, with higher rates among younger users. Not every blocker blocks analytics, and not every blocker user buys, so do not subtract 29.5 percent from your revenue. Measure it instead (see the reconciliation method below).
6. Tag failures and missing IDs
Customers who pay on a third-party page (PayPal, some buy now pay later providers) and never return to the thank-you page produce no purchase event. A theme update can also remove the data layer push. These failures cluster by payment method and by date, which is how you find them.
7. Duplicates, the cause that inflates GA4
Google’s documentation says that GA4 deduplicates purchase events with the same transaction ID, but only for the same user, so a reload of the confirmation page by the same visitor is safe. Two cases break this. A plugin that sends a different ID format (Magento’s increment ID from one integration and the entity ID from another) produces two transactions. And an empty string is treated as one ID, so GA4 will deduplicate every purchase that has transaction_id=”” into one, which silently undercounts.
Summary table
| Cause | Effect on GA4 | How to detect it | Fix or treatment |
|---|---|---|---|
| Unpaid, failed, cancelled orders | GA4 lower than raw admin total | Group store revenue by status | Filter store side to paid statuses |
| Refunds | GA4 higher than net store revenue | Compare gross to gross, then net to net | Send refund events with the original transaction ID |
| Time zone mismatch | Daily figures shift, monthly totals barely move | Compare hourly order counts | Align GA4 property time zone with the store, or convert in your query |
| Declined consent | Lower, roughly in line with decline rate | Consent platform acceptance rate | Advanced consent mode if eligible for modelling; otherwise report the gap |
| Ad blockers and privacy tools | Lower, skewed by device and audience age | Unmatched orders by browser and device | Server-side tagging, or accept and report a coverage rate |
| Tag failures | Lower, clustered by payment method or date | Unmatched orders by payment method | Fire purchase server-side or from a webhook |
| Duplicate or malformed IDs | Higher (duplicates) or lower (empty IDs) | GA4 transaction IDs that match nothing in the store | One integration, one ID format, never empty |
The reconciliation method: match orders, not totals
Comparing two monthly totals tells you the size of the gap but not its cause. Matching at the order level tells you both. Export GA4 transaction IDs with revenue for the period (the Transactions report, an Exploration, or the Data API), pull the same period from the store, and join on order number.
For WooCommerce with High-Performance Order Storage enabled, orders and refunds live in the same table, distinguished by the type column (shop_order or shop_order_refund), with refunds pointing at their order through parent_order_id:
-- Paid orders in the period, in UTC
SELECT o.id AS order_id,
o.status,
o.total_amount,
o.date_created_gmt,
o.payment_method,
COALESCE(r.refunded, 0) AS refunded
FROM wp_wc_orders o
LEFT JOIN (
SELECT parent_order_id, SUM(ABS(total_amount)) AS refunded
FROM wp_wc_orders
WHERE type = 'shop_order_refund'
GROUP BY parent_order_id
) r ON r.parent_order_id = o.id
WHERE o.type = 'shop_order'
AND o.status IN ('wc-processing', 'wc-completed', 'wc-refunded')
AND o.date_created_gmt >= '2026-10-01'
AND o.date_created_gmt < '2026-11-01';
On Magento, the equivalent is sales_order (use increment_id as the key if that is what your GA4 integration sends) with credit memos in sales_creditmemo.
The join produces three groups, and each one answers a different question.
- Store only. These are the orders GA4 never saw. Break them down by payment method, device, browser and country. A cluster on one gateway points to a tag failure. A spread across all gateways, heavier on desktop Firefox and Brave, points to blockers. A cluster in EU countries points to consent.
- Matched. Compare the revenue values. A consistent ratio usually means one side includes tax or shipping and the other does not. Pick a definition and send the same one to GA4.
- GA4 only. A healthy setup has close to none. Anything here is double counting, test orders, or an ID format that does not match the store.
Joining on order IDs from two systems is exactly where key mismatches drop rows without warning, for example when one side stores the ID as text with a prefix and the other as a number. QuantumLayers covers the checks worth running before trusting such a join in Joining Datasets Without Breaking Them.
Two traps that look like tracking problems
Thresholding. When Google signals is on and a report slices data into small groups, GA4 withholds rows. Google’s own help page says data thresholds are system defined and cannot be adjusted. A product-level or city-level revenue report that adds up to less than the total may be thresholded, not broken. The orange warning icon on the report tells you which.
Attribution windows. GA4 assigns revenue to the date of the purchase event, but channel reports can still shift for a few days as late data arrives. Reconcile complete months, a few days after month end.
What to report once the gap is understood
Use the store database for revenue, orders and average order value. Use GA4 for what it is good at: traffic sources, landing pages, funnels and on-site behaviour. Then publish the match rate next to every GA4 revenue figure so readers know how much of reality it covers. When the match rate drops from 82 to 64 percent in a week, you have a tracking incident, and you will know before anyone reallocates a budget because a channel “stopped converting”.
Once the GA4 export and the store query are scripted, the weekly match becomes a scheduled job instead of a spreadsheet chore.
Sources
- Google Analytics Help, [GA4] Behavioral modeling for consent mode
- Google Analytics Help, Measure ecommerce
- Google Analytics Help, [GA4] About data thresholds
- WooCommerce Developer Blog, High-Performance Order Storage: Database Schema
- GoDaddy Help, What are WooCommerce Order Statuses?
- WeltPixel, Refund Tracking on Magento 2: Credit Memos and What Reaches GA4
- Backlinko, Ad Blocker Usage and Demographic Statistics (GWI data)


