Top 30 Amazon Financial Analyst Interview Questions and Answers

Accrual accounting records revenue when it is earned and expenses when they are incurred, subject to the applicable accounting requirements, regardless of when cash changes hands.

Cash accounting records transactions when cash is received or paid.

For example, if a company delivers a service in March but receives payment in April, the revenue would generally be recognized in March under accrual accounting if the recognition criteria are met. Under cash accounting, it would be recorded when payment is received.

Assume a tax rate of (t), sufficient taxable income to realize the tax effect, and no other changes.

• Income Statement: Operating income decreases by $10. Net income decreases by 10(1-t).
• Cash Flow Statement: Net income starts lower by 10(1-t), but the additional $10 depreciation is added back because it is a non-cash expense. Cash flow from operations increases by 10t, reflecting the tax shield under these assumptions.
• Balance Sheet: Net PP&E decreases by $10. Cash increases by 10t from the tax effect, and retained earnings decrease by 10(1-t).

The balance sheet remains balanced.

Purchasing inventory does not immediately create an expense on the Income Statement.

The transaction generally increases inventory by $10 and decreases cash by $10, or increases accounts payable if the purchase is made on credit.

On the Cash Flow Statement, the purchase affects operating cash flow through the payment or working capital movement, depending on how the transaction is funded and recorded.

The expense is generally recognized in Cost of Goods Sold when the inventory is sold. If the inventory becomes obsolete or its recoverable value declines, a write-down may create an expense earlier.

Free Cash Flow (FCF) measures cash generated after capital expenditure.

A commonly used definition is:

EBITDA is earnings before interest, taxes, depreciation, and amortization. It is often used as an operating performance measure, but it does not account directly for capital expenditure, working capital requirements, or cash taxes.

For a capital-intensive business, EBITDA can therefore look healthy even when substantial cash is being consumed by infrastructure investment.

Net Working Capital can be defined as:

Negative working capital occurs when accounts payable exceed accounts receivable plus inventory under this definition.

In a high-velocity retail business, customers may pay quickly, inventory may turn rapidly, and suppliers may provide extended payment terms. The business can therefore receive cash before it needs to pay suppliers.

This can support operating cash flow and help fund business growth. However, negative working capital is not automatically healthy; its sustainability depends on the operating model and supplier relationships.

The Cash Conversion Cycle (CCC) measures how long a business's cash is tied up in its operating cycle.

Where:

• DIO Days Inventory Outstanding — the time inventory remains on hand.
• DSO Days Sales Outstanding — the time required to collect receivables.
• DPO Days Payable Outstanding — the time taken to pay suppliers.

A shorter or negative cycle can improve cash efficiency because cash is recovered more quickly relative to supplier payments.

An analyst should examine all three components rather than relying on the final number alone.

I would investigate four main areas:

I would quantify the contribution of each factor and distinguish temporary investment-related cash pressure from underlying operational deterioration.

Driver-based modeling forecasts financial outcomes using measurable operational inputs.

For example, a fulfillment labor expense model might use order volume, labor productivity, wage rates, and overtime assumptions to estimate total labor costs.

A simplified model could be:

Labor hours would then be estimated from the operational workload and productivity assumptions.

The key is to connect physical business activity to financial results rather than forecasting expenses solely from historical trends.

Driver-based forecasting is forward-looking. It estimates future financial performance using assumptions about business activity, such as volume, wage rates, utilization, and productivity.

Variance analysis is retrospective or diagnostic. It compares actual results with a budget or forecast to identify and explain differences.

For example, a driver-based model predicts future fulfillment costs based on expected order volume and productivity. Variance analysis investigates why actual fulfillment costs exceeded the budget last month.

Both methods are important, but they answer different questions.

Price-Volume-Mix (PVM) analysis breaks a financial variance into the effects of changes in selling prices, sales volumes, and the composition of products sold.

• Price variance: The effect of selling products at different prices from the baseline.
• Volume variance: The effect of selling a different total quantity.
• Mix variance: The effect of changes in the proportions of products sold.

For example, revenue might increase because selling prices rose, more units were sold, or customers purchased a greater proportion of higher-priced products.

I would calculate the relevant effects using a consistent baseline and then explain which business drivers caused the change.

I would begin by defining the investment decision and identifying the major cost and benefit drivers.

The model would include:

1. Volume assumptions: Expected orders, packages, and capacity utilization.
2. Operating costs: Labor, utilities, transportation, maintenance, and facility costs.
3. Capital expenditure: Construction, equipment, and technology investments.
4. Revenue or operational benefits: Incremental demand, faster delivery, and potential improvements in customer retention.
5. Cash flow: Initial investment, operating cash flows, and ongoing capital requirements.
6. Scenarios: Base, upside, and downside cases for volume, utilization, costs, and ramp-up time.

I would evaluate the investment using an appropriate financial framework, such as net present value, and test whether the expected benefits justify the investment under different assumptions.

The final recommendation should also consider delivery performance, operational feasibility, and customer impact.

I would examine the relationship between infrastructure investment, capacity utilization, revenue generation, and long-term cash flow.

Key inputs would include capital expenditure, available power capacity, server and accelerator utilization, expected customer demand, operating costs, depreciation, and hardware replacement cycles.

I would build a discounted cash flow model, evaluate the ramp-up in utilization, and test how changes in demand, pricing, and useful life affect returns.

I would also consider whether customer commitments and prepayments provide sufficient visibility into future demand, without assuming that commitments automatically eliminate investment risk.

Shortening the estimated useful life of servers generally accelerates depreciation expense over the asset's remaining depreciable life.

This can reduce reported operating income and net income in the near term. Because depreciation is a non-cash expense, it does not directly reduce pre-tax cash flow in the same way that a cash operating expense does.

The change may affect cash taxes and therefore after-tax cash flow, depending on the applicable tax treatment.

For cloud infrastructure, realistic useful-life assumptions also help align accounting costs with hardware obsolescence and replacement economics.

Based on the supplied research, candidates should be prepared to use:

In a case study, technical accuracy is only part of the evaluation. You should also make assumptions explicit, keep calculations auditable, and explain the business implications of your results.

I would start by validating the source data and confirming that actual and budget figures use consistent periods, categories, and definitions.

Next, I would use lookup functions to map transactions to the relevant categories, SUMIFS or PivotTables to aggregate the figures, and variance calculations to identify the largest differences.

I would then drill into the underlying transactions and operational drivers to determine the cause.

Finally, I would document the assumptions, validate the totals against the source records, and present the main findings with recommended actions.

I would first filter positive transactions and aggregate their amounts by calendar date. Then I would use a window function to calculate the rolling average.

WITH daily_totals AS (
    SELECT
        CAST(transaction_date AS DATE) AS txn_date,
        SUM(amount) AS daily_amount
    FROM transactions
    WHERE amount > 0
    GROUP BY CAST(transaction_date AS DATE)
)
SELECT
    txn_date,
    daily_amount,
    AVG(daily_amount) OVER (
        ORDER BY txn_date
        ROWS BETWEEN 2 PRECEDING AND CURRENT ROW
    ) AS rolling_avg
FROM daily_totals
ORDER BY txn_date;

This calculates the average of the current date and the previous two rows in the daily totals.

WHERE filters individual rows before aggregation. HAVING filters groups after aggregation.

For example:

SELECT
    department,
    SUM(expense) AS total_expense
FROM financial_data
WHERE fiscal_year = 2026
GROUP BY department
HAVING SUM(expense) > 100000;

Here, WHERE restricts the source rows to the specified fiscal year. GROUP BY calculates the total expense for each department, and HAVING returns only departments whose total expense exceeds 100,000.

This distinction is important when building financial reports and validating aggregated figures.

I would join the transaction table to itself using a common customer or transaction identifier, then group by the product pair and count how often each pair occurs.

A condition such as t1.product_id < t2.product_id prevents duplicate mirrored pairs, such as A–B and B–A.

A simplified example is:

SELECT
    t1.product_id AS product_1,
    t2.product_id AS product_2,
    COUNT(*) AS pair_count
FROM transactions t1
JOIN transactions t2
    ON t1.user_id = t2.user_id
   AND t1.product_id < t2.product_id
GROUP BY
    t1.product_id,
    t2.product_id
ORDER BY pair_count DESC
LIMIT 5;

The exact join key depends on the business definition. If the question asks for products purchased in the same order, I would join on an order or basket identifier rather than user ID alone.

In standard SQL grouping behavior, rows with NULL in the grouping column are placed into the same group.

For example, if some financial transactions have no assigned cost center, a query grouped by cost center can return a group representing those missing assignments.

This is important in financial reporting because unassigned records should not be silently ignored. I would investigate the missing values, determine whether they indicate incomplete mappings or genuine exceptions, and reconcile them before presenting the results.

CRaP stands for Can't Realize Any Profit. In the supplied research, it refers to products whose economics become unprofitable after relevant costs such as fulfillment, shipping, and payment processing are considered.

I would investigate four potential actions:

1. Renegotiate supplier costs: Seek a lower purchase price or improved vendor funding.
2. Optimize packaging and fulfillment: Reduce handling, storage, and shipping costs where feasible.
3. Change the selling format: Consider multipacks or bundles that improve the economics per shipment.
4. Review assortment decisions: If the product remains unprofitable and does not provide a justified strategic benefit, consider discontinuing it.

The final recommendation should account for contribution profit, customer demand, operational feasibility, and the product's broader role in the business.

I would first confirm that the cost-per-unit calculation and reporting period are consistent.

Then I would break the change into operational drivers, including:

• Order and package volume.
• Units processed per labor hour.
• Wage rates and overtime.
• Package weight and dimensions.
• Warehouse utilization and handling requirements.
• Transportation and delivery costs.
• Changes in service levels or product mix.

I would compare actual performance with the budget, quantify the largest contributors, and investigate whether the increase resulted from lower productivity, higher input costs, or a change in the operating model.

My recommendation would balance cost reduction with safety, delivery reliability, and customer experience.

I would start with the decision required, followed by the recommendation and its quantified financial impact.

The analysis would include the current business position, key assumptions, success metrics, major risks, and alternative scenarios. I would clearly distinguish verified data from estimates and explain why the recommended option is preferable.

I would then present the findings in concise, logical prose supported by financial tables and sensitivity analysis where necessary.

Amazon's narrative-memo culture, as described in the supplied research, emphasizes clear reasoning and evidence rather than relying on presentation formatting alone. The objective is to help decision-makers understand the recommendation, its trade-offs, and the actions required.