Skip to main content
Use voice-ordering analytics to review how often your team uses voice capture and where the results need attention. The report combines usage, editing, recognition speed and a daily activity chart for the selected shop.

​
Who can do this?

Owners and Managers can view the report. Staff cannot open these shop-wide voice analytics.

​
Where to do it

In the web app, open Overview, select View analytics, then open Voice orders.

​
Review voice ordering

  1. Check the shop in the shop picker.
  2. Choose a reporting period.
  3. Read the usage card and the displayed counts of voice orders and total orders.
  4. Review the cards for orders used without changes, manual edits and phrases that could not be matched to menu items.
  5. Check recognition time and microphone time to understand the time spent capturing orders.
  6. Hover over the daily chart to see voice-order counts and the available rate for orders used without changes.
Read the information notes beside the figures. The cards measure different parts of the ordering process; they are not interchangeable percentages that should add up to 100%.

​
What the figures tell you

Manual edits include changes to recognized items and removals. An unmatched phrase means recognition could not connect some spoken wording to a menu item. It does not mean that every part of the order failed. Recognition time measures successful attempts to interpret speech. The typical and slower-result values help distinguish normal processing from delays; they do not measure the whole checkout. Microphone time averages available recording durations for voice sessions submitted as orders, excluding abandoned or failed sessions. Voice-order counts reflect submitted voice sessions, not successful payments. The total-order count excludes draft baskets but can include cancelled or unpaid orders. A voice count can remain after its linked order is deleted, so do not use these figures to reconcile revenue. The edit rate includes recorded outcomes from abandoned sessions as well as submitted ones. The unmatched-phrase rate counts successful interpretation attempts with at least one unresolved phrase; it is not the percentage of individual words recognized incorrectly. Daily groups use UTC dates. Activity near midnight can therefore appear on a different day from your shop’s local reports. Retention cleanup removes speech content while keeping summary measurements, so older figures may remain after recordings and transcripts are unavailable. Cleanup also removes unmatched-phrase details while those attempts remain in the report’s total. The unmatched rate can therefore fall after cleanup without recognition improving. Avoid comparing this rate across periods with different amounts of retained speech data.

​
Platform and market notes

This report is in the web app and summarizes voice ordering used in the mobile app. It is available to the same roles in Great Britain and Germany. The longest voice-reporting window is 365 days. Selecting the maximum-history option uses this one-year limit; check the period shown on the report.

​
Troubleshooting

A dash means there is not enough information to calculate that rate, rather than a measured zero. If the chart has no voice orders, check the shop and period. Use the retry action if loading fails. Confirm your Owner or Manager role if access is restricted.