Read Testing Performance

Interpret product and campaign evidence, date ranges, ROAS, break-even context, and review states.

Before you begin

Confirm the active organization, selected date range, connected commerce and advertising sources, and the product or campaign you want to review. Testing metrics depend on synchronized evidence and SKU or campaign matching. A visible row does not guarantee that every economic input is complete.

How it works

Testing presents supported performance perspectives for combined products, Meta product evidence, and Google campaign evidence. The available tabs can change with the selected platform. Rows combine spend, sales, orders, matches, cost information, ROAS, break-even context, decision labels, reasons, and review state where those values are available.

Step-by-step

Open Testing in the correct organization.
Select the relevant date range and note that changing it changes the evidence set.
Choose the combined, Meta, or Google perspective supported by the current platform selection.
Search for the product SKU or campaign under review.
Read sales, spend, orders, and matching evidence before focusing on ROAS alone.
Compare reported ROAS with break-even context only when product economics are complete.
Read the decision reason and setup-health signals attached to the row.
Open supporting order or attribution evidence when available.
Record or update the review state after a human has assessed the evidence.
Export only after confirming filters, platform, and date range.

Check your result

You can explain which source, date range, orders, campaigns, products, and economic inputs produced the row. The review state reflects the organization’s follow-up, and any decision label is treated as evidence-supported guidance rather than an automatic instruction.

Common problems

ROAS is present but break-even is not: the product may be missing approved cost or other economics.

The product is absent: verify SKU matching, source connection, date range, and whether the selected platform supplies that view.

Meta and Google totals differ: they represent different supported evidence and attribution contexts; do not assume they should match.

An export looks incomplete: recheck active filters, segments, date range, and platform before exporting again.

Permissions and data notes

Testing requires module access and usable source data. Advertising and commerce figures can be delayed, attributed differently, or incomplete. Review sensitive performance exports according to organization policy and avoid presenting calculated evidence as final accounting data.