Cloud Kitchen — Delivery First
The situation
Seoul Sandwich was a delivery-first Korean sandwich concept launched from a cloud kitchen in Padi, Chennai in October 2023. The concept was built around Korean-influenced fillings — gochujang chicken, bulgogi, kimchi variants — in a format that worked for delivery: contained, portable, holding heat reasonably well, and photographable at a delivery price point.
The kitchen operated entirely on third-party delivery platforms. No dine-in, no direct ordering, no walk-in trade. Revenue was entirely Swiggy and Zomato dependent. The business ran for 15 months before closing in January 2025.
What was built
Menu design for delivery. The menu was designed for delivery physics from the start — not adapted from a dine-in menu. Fewer items (8 core SKUs at launch), structural integrity over 20–30 minutes of travel, price points calibrated to platform fee structures, photography that worked as a thumbnail at 200px. The final menu ran at 12 items, expanded after the first three months based on platform data.
Platform setup and listing optimisation. Both platforms set up from scratch. Platform ranking is partially driven by factors within operator control: opening hours, acceptance rate, cancellation rate, preparation time accuracy, listing completeness. All were managed actively from day one.
Packaging. Several iterations in the first two months before settling on the final system. Heat retention, structural integrity, presentation on opening, cost per unit that works within food cost budget. The initial packaging failed under real delivery conditions — 30 minutes in a delivery bag, mid-summer — before committing to volume orders.
Unit economics tracking. Weekly: food cost %, platform commission impact, AOV, order count, refund and cancellation rate, effective margin per order after platform fees. This data was the primary decision-making tool throughout.
What worked
Delivery menu logic. Designing for delivery physics from the start rather than adapting a dine-in menu was the right call. Items held well, photography worked at scale, and the menu was coherent rather than sprawling. Operators who start with too many items, or transfer a full dine-in menu to delivery unchanged, pay for it in preparation complexity, food cost, and rating inconsistency.
Platform listing management. Active management of preparation time accuracy, acceptance rates, and listing completeness produced measurable improvement in platform placement over the first three months. These factors are underestimated by most cloud kitchen operators who treat the platform as passive infrastructure.
Packaging iteration before volume commitment. Testing under actual delivery conditions before committing to 500 units avoided a more expensive mistake. The problem only appeared under real conditions, not in room-temperature tests.
What did not work
Platform commission impact on margin. At 20–30% platform commission combined with payment gateway fees, the effective margin per order on a mid-price delivery concept in Chennai is significantly thinner than initial projections suggest. The business model depends on volume, and at the order volumes achievable from a single cloud kitchen in a competitive category, sustainable margin was difficult.
Category competition. The Korean food category on Swiggy and Zomato in Chennai grew significantly over the same 15 months. More operators entering the category increased customer acquisition cost via discounting and compressed visibility windows for individual listings. The platform owns the customer relationship.
The dependency problem. 100% platform dependency means 100% exposure to policy changes, commission adjustments, algorithm changes, and competitive ranking dynamics. No direct customer relationship, no email list, no repeat customer mechanism outside platform favouriting. This is the structural weakness of the model that is difficult to resolve without direct ordering infrastructure.
What came from it
Closing was a decision, not a surprise. The data made the case clearly enough. The business was not going to reach sustainable unit economics at its current structure and scale without a level of capital investment in platform advertising that did not make financial sense.
The specific basis for the Cloud Kitchen Setup and Optimisation service: the decisions that determine whether the model is viable before significant capital is committed — platform setup, delivery menu design, unit economics modelling, packaging, and the platform dependency problem.
If you are building or running a cloud kitchen
Describe your specific situation in the booking form. I will tell you honestly whether what I know is useful to your problem.