How algorithms support our courier operations

The debate about algorithms and platform work matters, but it only works when it is grounded in how systems actually operate. In our transparency report, How Algorithms Support Couriers, we set out in detail how our tools support couriers from onboarding to delivery, pay, and account decisions. Here we focus on the five concerns we hear most often.

The thread through all five is the same principle that runs through everything we build: algorithms support, people decide. For any decision that materially affects a courier's access to work, income, or standing on the platform, a human reviews it before it takes effect.

No tracking after the run ends

The concern: GPS tracking and constant monitoring mean couriers are under continuous surveillance.

The reality: Location data is collected only while a run is active, and tracking stops the moment the run ends. It is used for routing, delivery coordination, and order matching, not for individual surveillance. The app does not direct couriers along any particular route.

When couriers clock out, the tracking turns off.

Algorithms flag, but humans decide

The concern: Automated systems can deactivate workers without human review or a path to appeal.

No system at Just Eat Takeaway.com automatically deactivates a courier. Fraud and compliance tools generate alerts; human operations teams investigate and decide. Every deactivation is made by a person, and a courier who believes a decision is wrong has the right to appeal, reviewed by a different agent from the one who made the original decision.

Algorithms flag. Humans decide

Courier safety is a priority

The concern: Speed-optimisation algorithms push couriers to ride dangerously.

Our delivery-time estimates are calibrated for urban delivery and are not designed to require or reward speeding. We reward couriers for being available and completing deliveries in high-demand areas, not for going faster. Couriers are free to decline any order, take a break, and log off whenever they want. Our systems are built around getting the delivery done as planned, not getting it done faster.

Zero tolerance for discrimination

The concern: Algorithms trained on historical data can replicate or amplify existing inequalities.

Assignment decisions are based on proximity, availability, and route efficiency, not on any characteristic ascribed to the courier. Protected-class attributes play no role in how orders are matched or how pay is calculated, and some systems deliberately avoid using individual courier history, relying on zone-level patterns instead. Customer ratings do not automatically trigger sanctions. Location and availability. That is how our system matches couriers to orders.

Equal and transparent pay

The concern: Personalised dynamic pricing means couriers doing the same work receive different pay.

Pay is not personalised. The same delivery, for the same partner and customer under the same conditions, is priced identically regardless of which courier receives it, and the full amount is always visible before acceptance. Where a courier waits beyond a set threshold or covers extra distance through no fault of their own, additional reimbursements apply.

Pay is consistent, transparent, and visible.

Why this matters

These are legitimate questions to ask of any platform. Our point is that the answers are specific, documented, and built on one principle: automated tools inform, and people decide. We publish this because at JET, transparency is a standard.

Read the full transparency report, How Algorithms Support Couriers, here.

Receive latest company news

Are you a journalist or do you work for a publication?
Sign up and request access to latest company news.

Request access

Receive Just Eat Takeaway.com news on your RSS reader.

Or subscribe through Atom URL manually