Diagnose LA completed-order drop and design experiment

On the week of 2025-08-25, completed orders in Los Angeles dropped 12% versus the week of 2025-08-18. Early indicators: sessions +4% WoW, order creations +1%, accept rate fell from 94% to 88%, p95 ETA rose 42→58 minutes, and cancellations with reason = 'no_courier' increased from 1.5% to 5.8%. Restaurant availability, pricing, and marketing spend were unchanged. Most of the decline occurred 18:00–21:00 local time.

Design a rigorous investigation and test plan that cleanly separates demand-side from supply-side causes and validates the true driver: A) Diagnosis: Lay out a funnel from session → order_created → accepted → picked_up → completed. Specify the exact cuts (by hour, ZIP, courier density, restaurant category, distance bands, surge level), sanity checks for tracking/config regressions, and the minimal set of plots/tables you would produce to prove whether supply is the bottleneck. What additional metrics would you compute, and what thresholds would convince you? B) Causal validation: If supply shortage during the dinner peak is your leading hypothesis, propose an experiment to fix it (e.g., targeted courier incentives or pricing changes). Define: experimental unit and randomization (e.g., ZIP clusters within LA with interference mitigation), primary and guardrail metrics, success criteria, power/MDE assumptions, sample size, duration, and a ramp plan. Address seasonality and spillovers explicitly. C) If an RCT is infeasible, specify a quasi-experimental alternative (e.g., difference-in-differences or synthetic control using other SoCal metros), including pre-trend checks and robustness tests. Provide a concrete analysis checklist for either path so that another analyst could execute it without ambiguity.