Skip to main content
Promotion details combine operational counts with an estimated impact card. Use redemptions and discount cost as direct records; treat the impact estimate as analysis that may be unavailable or within normal variation.

​
Who can do this?

Owners and Managers can perform this flow. Staff may see a link in some navigation states, but the promotion API does not grant administration access.

​
Where to do it

Use Promotions in the Management web app for the labels in this guide. Equivalent promotion screens also exist in the Management mobile app. POS actions happen in the cart; guest-facing effects appear in the guest web store.

​
Steps

  1. Open a promotion and read Redemptions against its total cap, or note that no cap is configured.
  2. Review the minimum-spend rule and discount cost so you understand the campaign’s operational constraints and recorded cost.
  3. In Promotion impact, choose the available metric, such as net revenue or order count, and read the comparison window and peer-store context.
  4. Check the effect band. Within normal variation means the model did not find a measurable change outside its noise band.
  5. Read any measurement explanation: insufficient history, low accuracy, or an overlapping change can prevent a reliable estimate. The first evaluation runs after launch rather than instantly.

​
Platform notes

Web and mobile use the same promotion and authorization rules from the API. The guest app lists only guest-visible, currently applicable offers; voucher links can carry a code into the basket. A control shown on one platform never expands the server-enforced role.

​
Market notes

The flow is the same for GB and Germany. Money is displayed in the store currency, and schedules are evaluated in the store’s local time zone. Check the selected store, currency, date, and time rather than relying on country assumptions. This guide reflects the product on 17 July 2026. The API-enforced role and saved promotion state are authoritative; local previews are guidance and the server validates the rules again when the change is persisted.

​
Troubleshooting

  • Do not interpret Not yet evaluated as zero impact. Revisit after the overnight evaluation following launch.
  • Redemption counts show use, not causality. If the impact card says the result cannot be measured or is within noise, do not present it as a proven uplift or decline.