A Zomato-style team sees a regional drop in qualified restaurant selection rate. How would you isolate the cause?

Problem Statement Description

Product context: Zomato is an Indian food-tech and local commerce company; its products include restaurant discovery, reviews, food delivery, dining, ads, restaurant partner tools, and quick commerce through Blinkit.

You are investigating an unexpected regional decline in the qualified restaurant selection rate for a Zomato-like restaurant discovery experience. A “qualified selection” should be treated as a user choosing a restaurant that appears to satisfy their intent and constraints, such as cuisine, location, availability, ratings, dietary needs, price range, delivery/bookability, or trust signals.

The issue is concentrated in one region and affects urban diners, especially users with dietary constraints who rely on accurate filters, menus, reviews, and restaurant metadata before ordering or booking. The drop may have implications for customer trust, restaurant visibility, sponsored placements, and downstream conversion, but the cause is not yet known.

Your task is to describe how you would isolate the cause of the anomaly. Focus on how you would frame the metric movement, validate that the drop is real, segment the problem, generate hypotheses, gather evidence, and decide what to investigate or mitigate first.

The experience should consider:

- Clear definition of qualified restaurant selection rate, including numerator, denominator, user action, and eligibility rules

- Regional and temporal framing, including when the drop started, whether it is sudden or gradual, and comparison to control regions

- Segmentation by user cohort, dietary constraint, cuisine, device, app version, traffic source, time of day, and restaurant category

- Instrumentation and data-quality checks, including event logging, taxonomy changes, filter behavior, ranking changes, and missing restaurant attributes

- Marketplace factors such as restaurant availability, menu freshness, delivery radius, table inventory, closures, pricing, and sponsored ranking pressure

- Trust and content signals such as review quality, rating changes, photo/menu accuracy, and moderation or recommendation changes

- Evidence needed to distinguish product, data, supply, demand, ranking, and external-market causes

- Immediate mitigation, monitoring, stakeholder communication, and prevention mechanisms once the cause is narrowed

The goal is to demonstrate a structured root-cause analysis approach that can separate measurement artifacts from real user experience degradation, identify the most likely drivers, and guide the team toward confident next steps without jumping prematurely to a solution.