Rescue Radar: Food Delivery Post-Cancellation Offers

Food delivery platforms have spent a decade optimizing the forward path from order to doorstep, and almost no time on the failure path. A canceled order is already cooked, already paid for by the restaurant, and already sitting in a rider bag. The operational conditions for a salvage product have never been more favorable.

A sustainability-minded user cohort is now commercially meaningful. Some users will choose a discounted rescue over a full-price fresh order, not despite the waste story but because of it. That shift makes rescue a viable product rather than a marketing line.

Every canceled order today is a triple loss. The restaurant eats the food cost, the platform eats the refund, and the rider is idle. No surface exists to turn that loss into a second sale.

The user-side problem is discovery under a sharp constraint. A rescued dish has a window of roughly three to five minutes and a radius of about five hundred meters. Any interface that relies on search or browse will miss it. Discovery has to come to the user in real time.

The first bet is that a rescue surface should not live inside the normal delivery flow. A separate, map-first, time-pressured surface makes the short window and close radius feel like the fun of the product rather than its friction.

The second bet is that framing matters as much as pricing. A discounted dish must read as "a neighbor just freed up this order" and not as "someone rejected this food." The leaderboard framing around Waste Warriors and CO2 saved turns a salvage economy into a status game, which is what actually produces returning users.

A feed-first discovery surface, closer in feel to a short-form video feed than to a restaurant grid. A live Rescue Radar map shows nearby rescue offers appearing in real time. A ninety-second hold timer runs once a user secures a deal, which creates the urgency the product depends on without pushing the window so short that the user cannot act.

The Pulse ticker makes the marketplace legible as a live system, showing rescues happening right now against food waste prevented over the last hour. A gamified checkout ranks users on cumulative CO2 saved, which turns retention into a leaderboard rather than a discount coupon.

Global state lives in a RescueContext, so offers, cart, and hold timers stay in sync across the feed, the map, and the checkout without routing complexity.

I did not integrate a real delivery backend. The point of the prototype was to validate the surface and the framing, not the operations. A real integration would require a rider re-route planner, which is a separate product.

I chose the gamified Waste Warriors frame over a pure discount frame. A pure discount would likely convert faster in the short term, but it would train users to treat rescue as a price event rather than a habit, which undercuts the reason the feature exists.

The prototype uses a cached library of offer copy rather than an LLM-in-the-loop. A live LLM would be slower and more expensive on the notification path, and for a salvage product the notification has to land inside a narrow window.

Rescue Radar converts a refund line item into a revenue line item. Even a modest rescue rate recovers margin that would otherwise be zero, and it protects the restaurant relationship that cancellations damage the most.

The sustainability framing is not a marketing layer. The CO2 leaderboard is a retention mechanic that is difficult for a competitor to copy, because copying it requires rebuilding the upstream waste-prevention story rather than just adding a badge.

The right launch region for this feature is a dense metro with a high cancellation rate and a sustainability-aware user base. A prototype like this is what lets a growth team see the surface clearly before committing to a launch geography.