Data Quality
Part of Ecommerce data quality
Detecting missing currencies and order values
Check purchase events for usable currency and value fields, distinguish genuine zeroes from missing data, and trace exceptions to orders.
Enable debug mode in the GA4 implementation and inspect purchase events in real time. For each purchase, check the event-level currency and value fields and inspect the items array.
Find incomplete purchase events
For each event, record whether currency and value are present and non-empty. Classify absent, empty, wrongly typed and unexpected fields separately; a quoted amount such as "49.00" is a string and needs correction.
Compare currency with the order's checkout currency; treat a mismatch as unexpected. Check that value is usable as an amount, and compare it with the chosen order amount only after defining what that amount represents.
A zero value may reflect a legitimate transaction under the store's rules, so inspect the underlying order before calling it an error. Show the number of purchase events inspected alongside the number with unusable fields, and include incomplete purchases in the inspected total.
Compare amounts using the same definition
Choose the amount to compare before looking for mismatches. A commerce report may use another sales definition, so comparing event value directly with a tax-inclusive order total can create a false discrepancy.
Where a usable order identifier exists, inspect the matching order's checkout currency, item amounts, discounts, tax and shipping alongside the sent payload. Check whether a blank mapping, a currency symbol inside value, or an amount from the wrong checkout stage explains the exception.
For multi-currency orders, keep the checkout currency with the original amount.
Check item revenue separately
Google Analytics uses purchase event-level value and currency for event-level revenue metrics. Item revenue depends on item price and quantity, with event-level currency relevant to its monetary reporting. If one revenue view is present and the other is missing, inspect the corresponding part of the payload before changing a dashboard calculation.
After correcting a mapping, inspect an approved test event and allow the report to process before assessing its totals. Mark the affected period. Do not silently replace missing historical payloads with estimates.
