Should a food-delivery app launch a subscription?

The prompt

A food-delivery app is considering a monthly subscription that gives free delivery on all orders. Leadership asks whether it is a good idea. How would you evaluate it, and how would you test it before a full launch?

Extra context you can assume

• Today every order pays a delivery fee, which is a meaningful part of revenue per order.
• Some customers order often; many order once or twice a month.
• The app can run experiments on a random subset of users.

A strong answer usually covers

• State the goal: more profit, higher frequency, or better retention?
• Model the economics: lost fee revenue vs extra orders, subscription fees and who self-selects into subscribing.
• Identify the metrics: order frequency, margin per user, retention, cannibalisation.
• Design an experiment: unit of randomisation, duration, success and guardrail metrics.
• Discuss risks and a decision rule.

This scenario is fictional and made up for practice.