Allocate Support Cost and Diagnose Decline

Assume the role of the analytics partner supporting the Customer Support team at a food-delivery company. The data available to you consists of three tables: agents (agent_id, monthly_salary, productive_hours, tenure, locale), tickets (ticket_id, created_at, resolved_at, issue_type, severity, channel, order_id, merchant_id, courier_id, csat, reopened_flag, escalated_flag), and ticket_touches (ticket_id, agent_id, minutes_spent, action_ts).

1. Propose a method for distributing salary cost or performance-linked pay across support agents fairly, based on completed complaint tickets. Describe how your approach would account for ticket difficulty, several agents handling the same ticket, quality, and the risk of gaming.

2. A key support KPI takes an abrupt downturn. Walk through how you would figure out whether instrumentation changes, seasonality, queue mix, staffing, or a genuine operational problem is responsible.

3. Complaints are increasing as a result of merchants making the wrong order, couriers collecting the wrong bag, or couriers dropping the order at the wrong customer. Which metrics would you monitor, how would you pinpoint the dominant driver, and which preventive measures would you put forward?

Overview: The question assesses operational analytics skills — cost allocation and attribution for support work, root-cause analysis of KPI regressions, and metric design for identifying and preventing complaint drivers.

Read the complete analytics engineer interview experience that this question was drawn from.