Diagnose P&L Variance in Grocery Delivery

Business Context

You’re the analytics lead supporting Uber Eats Grocery in the US (≈ 6M monthly active customers, 120K active couriers, operations in 40 metros). The business is in a profitability push: leadership committed to improving Contribution Margin while maintaining on-time delivery SLAs and customer retention.

At the end of January, Finance flags that the Grocery line missed plan materially. The GM asks you to “analyze the P&L, explain the variance, and tell me what we should do in the next 2 weeks.” You have access to both accounting P&L outputs and operational event-level data.

Metric Scenario

Finance provides the following simplified P&L (USD, January actual vs December actual vs January plan):

Line Item	Dec Actual	Jan Plan	Jan Actual
Orders	18.0M	19.5M	19.0M
Gross Bookings (GB)	1,440M	1,560M	1,482M
Net Revenue (NR)	216M	234M	215M
Variable Costs (VC)	153M	160M	170M
Contribution Margin (CM = NR - VC)	63M	74M	45M

Stakeholders disagree on the “why”:

Marketing claims promos were flat and blames courier incentives.
Ops claims weather drove longer trips and higher cancellations.
Product suspects a January checkout change reduced basket size and increased support contacts.

You need to reconcile the P&L with product/ops metrics, identify the biggest drivers, and propose actions with clear expected impact.

Available Data

You can query a warehouse (Snowflake) with these sources:

Table / Log	What it contains	Grain
`finance_pnl_daily`	daily NR, VC, promo spend, courier incentives, support cost allocations, chargebacks	day × metro × line_item
`orders`	order_id, user_id, store_id, metro, created_at, delivered_at, status, cancellation_reason	order
`pricing_promos`	order_id, promo_type, promo_amount, funded_by (company/merchant), fees, taxes	order
`courier_payments`	order_id, base_pay, surge, incentive, distance_km, time_minutes	order
`app_events`	checkout funnel events, payment failures, substitutions accepted, tip edits	event
`support_tickets`	ticket_id, order_id, issue_type, handle_time, refund_amount	ticket

What You Must Produce

Define the primary KPI you will use to evaluate P&L performance (not just “profit”), and explain why it’s the right metric for this business.
Compute and explain variances for January vs Plan and vs December using a structured approach (e.g., price/volume/mix, rate/efficiency). Make clear which variances are favorable/unfavorable.
Decompose the KPI into actionable drivers that map to product and ops levers (e.g., basket size, take rate, courier cost per order, cancellation rate, refunds).
Propose 3–5 hypotheses for the CM miss and how you would validate each using the available tables (include at least one hypothesis that reconciles accounting allocations with operational reality).
Recommend 3 concrete actions you would take in the next two weeks, with the expected direction of impact on the KPI and at least one guardrail metric per action.

Constraints:

You have 5 business days to deliver a readout for the monthly business review.
Any recommendation must not degrade on-time delivery by more than 1 pp or increase customer churn.

Business Context

Metric Scenario

Finance provides the following simplified P&L (USD, January actual vs December actual vs January plan):

Line Item	Dec Actual	Jan Plan	Jan Actual
Orders	18.0M	19.5M	19.0M
Gross Bookings (GB)	1,440M	1,560M	1,482M
Net Revenue (NR)	216M	234M	215M
Variable Costs (VC)	153M	160M	170M
Contribution Margin (CM = NR - VC)	63M	74M	45M

Stakeholders disagree on the “why”:

Marketing claims promos were flat and blames courier incentives.
Ops claims weather drove longer trips and higher cancellations.
Product suspects a January checkout change reduced basket size and increased support contacts.

You need to reconcile the P&L with product/ops metrics, identify the biggest drivers, and propose actions with clear expected impact.

Available Data

You can query a warehouse (Snowflake) with these sources:

Table / Log	What it contains	Grain
`finance_pnl_daily`	daily NR, VC, promo spend, courier incentives, support cost allocations, chargebacks	day × metro × line_item
`orders`	order_id, user_id, store_id, metro, created_at, delivered_at, status, cancellation_reason	order
`pricing_promos`	order_id, promo_type, promo_amount, funded_by (company/merchant), fees, taxes	order
`courier_payments`	order_id, base_pay, surge, incentive, distance_km, time_minutes	order
`app_events`	checkout funnel events, payment failures, substitutions accepted, tip edits	event
`support_tickets`	ticket_id, order_id, issue_type, handle_time, refund_amount	ticket

What You Must Produce

Define the primary KPI you will use to evaluate P&L performance (not just “profit”), and explain why it’s the right metric for this business.
Compute and explain variances for January vs Plan and vs December using a structured approach (e.g., price/volume/mix, rate/efficiency). Make clear which variances are favorable/unfavorable.
Decompose the KPI into actionable drivers that map to product and ops levers (e.g., basket size, take rate, courier cost per order, cancellation rate, refunds).
Propose 3–5 hypotheses for the CM miss and how you would validate each using the available tables (include at least one hypothesis that reconciles accounting allocations with operational reality).
Recommend 3 concrete actions you would take in the next two weeks, with the expected direction of impact on the KPI and at least one guardrail metric per action.

Constraints:

You have 5 business days to deliver a readout for the monthly business review.
Any recommendation must not degrade on-time delivery by more than 1 pp or increase customer churn.

Business Context

Metric Scenario

Finance provides the following simplified P&L (USD, January actual vs December actual vs January plan):

Line Item	Dec Actual	Jan Plan	Jan Actual
Orders	18.0M	19.5M	19.0M
Gross Bookings (GB)	1,440M	1,560M	1,482M
Net Revenue (NR)	216M	234M	215M
Variable Costs (VC)	153M	160M	170M
Contribution Margin (CM = NR - VC)	63M	74M	45M

Stakeholders disagree on the “why”:

Marketing claims promos were flat and blames courier incentives.
Ops claims weather drove longer trips and higher cancellations.
Product suspects a January checkout change reduced basket size and increased support contacts.

You need to reconcile the P&L with product/ops metrics, identify the biggest drivers, and propose actions with clear expected impact.

Available Data

You can query a warehouse (Snowflake) with these sources:

Table / Log	What it contains	Grain
`finance_pnl_daily`	daily NR, VC, promo spend, courier incentives, support cost allocations, chargebacks	day × metro × line_item
`orders`	order_id, user_id, store_id, metro, created_at, delivered_at, status, cancellation_reason	order
`pricing_promos`	order_id, promo_type, promo_amount, funded_by (company/merchant), fees, taxes	order
`courier_payments`	order_id, base_pay, surge, incentive, distance_km, time_minutes	order
`app_events`	checkout funnel events, payment failures, substitutions accepted, tip edits	event
`support_tickets`	ticket_id, order_id, issue_type, handle_time, refund_amount	ticket

What You Must Produce

Define the primary KPI you will use to evaluate P&L performance (not just “profit”), and explain why it’s the right metric for this business.
Compute and explain variances for January vs Plan and vs December using a structured approach (e.g., price/volume/mix, rate/efficiency). Make clear which variances are favorable/unfavorable.
Decompose the KPI into actionable drivers that map to product and ops levers (e.g., basket size, take rate, courier cost per order, cancellation rate, refunds).
Propose 3–5 hypotheses for the CM miss and how you would validate each using the available tables (include at least one hypothesis that reconciles accounting allocations with operational reality).
Recommend 3 concrete actions you would take in the next two weeks, with the expected direction of impact on the KPI and at least one guardrail metric per action.

Constraints:

You have 5 business days to deliver a readout for the monthly business review.
Any recommendation must not degrade on-time delivery by more than 1 pp or increase customer churn.

Business Context

Metric Scenario

Finance provides the following simplified P&L (USD, January actual vs December actual vs January plan):

Line Item	Dec Actual	Jan Plan	Jan Actual
Orders	18.0M	19.5M	19.0M
Gross Bookings (GB)	1,440M	1,560M	1,482M
Net Revenue (NR)	216M	234M	215M
Variable Costs (VC)	153M	160M	170M
Contribution Margin (CM = NR - VC)	63M	74M	45M

Stakeholders disagree on the “why”:

Marketing claims promos were flat and blames courier incentives.
Ops claims weather drove longer trips and higher cancellations.
Product suspects a January checkout change reduced basket size and increased support contacts.

You need to reconcile the P&L with product/ops metrics, identify the biggest drivers, and propose actions with clear expected impact.

Available Data

You can query a warehouse (Snowflake) with these sources:

Table / Log	What it contains	Grain
`finance_pnl_daily`	daily NR, VC, promo spend, courier incentives, support cost allocations, chargebacks	day × metro × line_item
`orders`	order_id, user_id, store_id, metro, created_at, delivered_at, status, cancellation_reason	order
`pricing_promos`	order_id, promo_type, promo_amount, funded_by (company/merchant), fees, taxes	order
`courier_payments`	order_id, base_pay, surge, incentive, distance_km, time_minutes	order
`app_events`	checkout funnel events, payment failures, substitutions accepted, tip edits	event
`support_tickets`	ticket_id, order_id, issue_type, handle_time, refund_amount	ticket

What You Must Produce

Define the primary KPI you will use to evaluate P&L performance (not just “profit”), and explain why it’s the right metric for this business.
Compute and explain variances for January vs Plan and vs December using a structured approach (e.g., price/volume/mix, rate/efficiency). Make clear which variances are favorable/unfavorable.
Decompose the KPI into actionable drivers that map to product and ops levers (e.g., basket size, take rate, courier cost per order, cancellation rate, refunds).
Propose 3–5 hypotheses for the CM miss and how you would validate each using the available tables (include at least one hypothesis that reconciles accounting allocations with operational reality).
Recommend 3 concrete actions you would take in the next two weeks, with the expected direction of impact on the KPI and at least one guardrail metric per action.

Constraints:

You have 5 business days to deliver a readout for the monthly business review.
Any recommendation must not degrade on-time delivery by more than 1 pp or increase customer churn.

Interview Guides

Business Context

Metric Scenario

Available Data

What You Must Produce

Diagnose P&L Variance in Grocery Delivery

Business Context

Metric Scenario

Available Data

What You Must Produce

Your Answer

Diagnose P&L Variance in Grocery Delivery

Business Context

Metric Scenario

Available Data

What You Must Produce

Diagnose P&L Variance in Grocery Delivery

Business Context

Metric Scenario

Available Data

What You Must Produce

Your Answer