In this guide

The purchase count matches the order system, but reported revenue is a hundred times too high. A common explanation is a mismatch between minor currency units and major units: an amount stored as 12,500 cents is sent as 12,500 dollars.

Value errors can distort reporting and any optimization that uses the value signal. Counting the right number of purchases is only part of the QA.

Use orders with known arithmetic

Prepare approved test cases for one item, several quantities, a discount, shipping, tax, and a refund where supported. Write the expected amounts before inspecting the event.

For a simple example, two items priced at $30 each with a $10 order discount produce $50 in merchandise value. Shipping and tax should be handled according to the destination’s event definition, not added or removed by guesswork.

Google’s GA4 ecommerce documentation describes value, currency, item data, and refund events. Use the specification of every receiving system because the same field name can have different inclusion rules.

Trace the amount through the pipeline

Compare the order record, data layer or source payload, outgoing event, and processed report. Identify the first place the value changes unexpectedly.

Check whether a decimal string is parsed correctly, whether thousands separators are accepted, and whether quantity is applied twice. Confirm that a fallback value does not replace a valid zero or discounted amount.

If a tag template transforms the value, record that transformation. Hidden arithmetic in several systems is difficult to reconcile.

Preserve the transaction currency

Send the required currency with the event value. A symbol displayed on the page is not enough for a machine-readable event.

For multi-currency sites, test each supported currency and distinguish the original transaction amount from any reporting conversion. Do not convert the amount yourself and then label it with the original currency.

Document the reporting currency and the basis of any exchange-rate conversion when comparing totals across systems.

Check identity and corrections

Use the transaction identifier as required by the destination’s duplicate-handling design. Refreshing a confirmation page should not create another independent sale.

A partial refund needs a different test from a canceled order. Confirm how the source and receiving platform represent those corrections, including whether the original purchase remains visible.

The server-side duplicate-conversion guide covers identity across collection paths. Value QA adds the question of whether the same correctly identified event carries the right amount.

Monitor after release

Review a sample of real, access-controlled transactions after the implementation goes live. Compare aggregate totals and inspect exceptions without exposing customer details in shared reports.

Set alerts for implausible changes in average order value, missing currency, and unusual concentrations at a default amount. Investigate before assuming the campaign suddenly attracted much more valuable customers.

The acceptance record should connect a known order to the correct event and processed value. That makes the result useful for both reporting and campaign decisions.

AdOpsNow Editorial

Examples use illustrative data. See our editorial standards or report a correction.