An order export includes tests and unpaid orders: define the sales population

Decide whether the review concerns orders placed, customer payments, fulfilled sales or accounting revenue. Those populations can differ. An unpaid order may still represent a genuine customer commitment, while an unconverted draft may remain only a proposed transaction. Keep the status fields needed to distinguish them before summing an export.

Shopify explains that test orders are excluded from its sales reports but can appear in order exports. Its sales-report guidance includes several open, pending or canceled order states and converted drafts. A raw export total therefore need not match the displayed sales report. Apply documented filters rather than assuming one source must be wrong.

Use the platform's test indicator, payment mode and approved test-order register where available. A customer name containing test is a clue but not a complete control. Employees can place genuine paid purchases, and a realistic test name can look like a customer. Verify whether real money moved and whether goods actually shipped before correcting a record.

Keep test transactions out of the relevant commercial performance population while retaining enough evidence to reconcile any real fees, refunds or stock movements they caused. If a live payment was used for checkout testing, its bank and processor events still need an explanation. Excluding the order from marketing analysis does not make the cash movement disappear.

Review integration behavior in test mode and live mode separately. A connector may receive a record that the storefront's own dashboard excludes. Use the connector's supported exclusion rules and confirm their effect on an actual controlled test.

Identify drafts that became orders, drafts still awaiting customer acceptance and completed orders awaiting collection. Preserve the conversion reference so the draft and final order are not counted twice. For manually paid orders, obtain receipt evidence rather than assuming the absence of processor settlement means the sale is invalid.

For a hypothetical export with $12,000 of order totals, $800 of confirmed test orders and $1,200 of unconverted drafts, the remaining order population is $10,000 before other differences. That is not automatically accounting revenue or collected cash. Review fulfillment, returns and the applicable recognition policy before using the amount in financial statements.

Save the original export, the filtered version and a count-and-value bridge for excluded records. Document the exact fields and values used. Avoid deleting source rows from the only copy or manually removing orders without a reason. The next reviewer should be able to reproduce the total and inspect a disputed inclusion.

Compare the result with the appropriate report using the same period and metric. Investigate residual differences such as later edits, refunds or grouping behavior. Review the filter when the platform or connector changes its fields. A reliable sales population is one whose inclusions and exclusions can be explained at the order level, with real cash and inventory events still accounted for.