Prove one city profitable first

The real moat in on-demand is not early footprint, it is proving one city can reliably make money before copying that playbook elsewhere. Homejoy expanded city count before fixing customer acquisition payback and repeat usage, while Instacart and later DoorDash won by tightening the economics of a repeat order workflow, then using that density to spread fixed costs and support broader expansion.

• Homejoy’s problem was structural. Cleaning had many one-off jobs, longer payback than a single order, and high disintermediation risk once a cleaner and customer met. That made each new city a new cash burn center, not a scaled copy of a proven market.
  1 sacra
• Instacart’s early model also lost money on deliveries, but it improved by using existing grocery stores instead of building warehouses, then layering on higher-margin retailer software and ads. That is what redoing unit economics looks like in practice, keep the order flow, remove avoidable cost, add better monetization.
  2 sacra
• Across on-demand, density changes everything. More orders in one zone mean shorter courier idle time, faster delivery, lower cost per order, and less paid marketing pressure. Research on Rappi shows the same pattern, concentrated markets and higher frequency create the conditions for profitability, then expansion.
  3 sacra

The next winners in delivery and local commerce will keep shrinking toward markets and categories where repeat behavior is strongest, then expand once the playbook is mechanical. Going abroad comes after a company knows exactly which customer cohort, basket size, take rate, and delivery pattern produce durable contribution margin.