On Dispatch, especially during peak hours, it is easy for items that need action to slip through the cracks. Alerts aims to solve this problem and increase service levels by highlighting these issues and generating buttons to take action quickly and easily.
On a repeating interval, the server searches for orders, shifts, or metrics that match the above triggers. If a match is found, it will check if an alert has already been created for the scenario. If not, the details of the problem, as well as the actions to solve the problem, are generated and sent to Dispatch to be displayed.
In Version 1 of alerts, the system will pick up on 5 scenarios:
| name | trigger | action |
|---|---|---|
order_ct |
current_time is 15+ minutes past when an order is marked as ready OR current_time is 15+ minutes past the estimated ready time based on the restaurant's prep_time |
call restaurant, apologize, give updated ETA |
order_unacknowledged |
order was assigned 5+ minutes ago and not yet accepted |
call runner to confirm, reassign if needed |
runner_ncns |
block started 15+ mins ago but runner has not checked in |
call runner - if no response, push block back and unassign |
restaurant_cancel |
high-volume restaurant cancels more than 25% of their daily orders |
call the restaurant to understand why they are canceling, prioritize them as necessary |
territory_high_drh |
current territory drh rises above the average target_drh for all retreat points in the territory |
generate a list of runners shifted later or not working, ask them to start early or pick up a block |
Based on existing ideas as well as feedback from the team, there will be more situations added to the list. Some planned alerts include:
- Runner moving away from next task
- Outdated runner location ping
- Runner out of zone
- Order has high TDT