Complaints increased for instant grocery substitution and availability after a release. How would you investigate?

Problem Statement Description

Blinkit has recently shipped a release affecting the instant grocery ordering experience, specifically around product availability and substitutions when an ordered item is out of stock. Soon after the release, customer complaints increased, with premium quick-commerce users reporting issues such as unavailable items, unwanted substitutions, missed substitutions, or confusion about what would actually arrive in their order.

You are asked to investigate this as a root-cause analysis problem. The focus is not to redesign the substitution flow immediately, but to structure how you would confirm the anomaly, isolate where it is happening, identify likely causes, and decide what actions are needed to protect customer trust and on-time fulfilled orders.

Assume the system involves customer-facing availability signals, dark-store inventory data, picker workflows, substitution rules, delivery SLAs, and customer support complaint channels. The release may have changed one or more parts of this chain, and the issue could be product, operational, data, or instrumentation-related.

The investigation should consider:

- How you would define and validate the complaint increase, including baseline period, complaint rate denominator, and affected complaint categories

- How you would segment the issue by city, dark store, SKU/category, customer cohort, app version, order value, time of day, and substitution type

- How you would check whether the release caused a real customer experience issue versus a tracking, tagging, or support-classification change

- What hypotheses you would explore across inventory accuracy, availability display, substitution recommendation logic, picker execution, customer consent, and delivery timing

- What data sources you would use, such as order events, inventory snapshots, picker actions, substitution acceptance/rejection, refunds, cancellations, and support tickets

- How you would prioritize severity based on customer impact, premium-user trust, fulfillment success, refunds, repeat orders, and operational load

- What short-term mitigations, rollback considerations, and monitoring you would expect while the root cause is being confirmed

- How you would prevent recurrence through better release checks, alerting, experimentation guardrails, and operational feedback loops

Your goal is to demonstrate a clear, structured RCA approach that connects customer complaints to the underlying product and operations workflow, separates signal from noise, and leads to evidence-based next steps without jumping prematurely to a solution.