← Capital One Interview Insights

Capital One·Data Scientist·Technical Phone Screen·Senior

Senior
Jul 2026

Summary

Capital One data scientist interview that went deep into subscription unit economics and insurance math. Six parts back to back, some straightforward, some genuinely tricky once churn and fixed costs entered the picture. Left feeling okay about the first half and less sure about the back half.

Questions Asked (6)

Q1

A subscription service charges $40/month with the first 3 months free, a $25/month variable service cost, $35 one-time installation, and $120 variable marketing/overhead per customer. The contract is 15 months with no churn. What is the net value per new customer?

Pricing & MonetizationProduct Analytics & Metrics
Author's notes

Pretty mechanical once you lay it out.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Calculate total revenue over the 15-month contract, accounting for the 3 free months, then subtract all variable costs (service, marketing/overhead) and the one-time installation fee. Present the net value per customer as a single figure, and briefly explain the key assumptions and their impact.

Pro tip: Clarify that the $25/month variable service cost and $120 marketing/overhead are per customer and incurred monthly, while installation is one-time. Also note that the first 3 months are free, so revenue is only for 12 months, but costs may still apply during free months—confirm with the interviewer if unsure.

1. Identify revenue streams and timing

Determine that the subscription fee is $40/month for 12 paid months (since first 3 are free) over a 15-month contract. No other revenue sources are mentioned.

2. Calculate total revenue

Multiply $40 by 12 months to get $480 total revenue per customer.

3. Identify and sum variable costs

List monthly variable costs: $25 service cost and $120 marketing/overhead. Sum to $145 per month. Multiply by 15 months (assuming costs incurred every month, including free months) to get $2,175.

4. Add one-time costs

Include the $35 installation fee as a one-time cost.

5. Compute net value

Subtract total costs ($2,175 + $35 = $2,210) from total revenue ($480) to get net value = -$1,730 per customer. Clearly state the negative result and discuss implications.

Key Points to Mention

  • Revenue is only for 12 months due to 3 free months.
  • Variable costs (service and marketing/overhead) are incurred monthly for all 15 months, including free months.
  • Installation fee is a one-time cost.
  • Net value is negative, indicating a loss per customer under these assumptions.
  • Assumptions about cost timing (e.g., whether marketing/overhead applies during free months) should be stated and can be adjusted if interviewer provides clarification.
  • The result highlights the importance of customer lifetime value and potential need for cost reduction or pricing changes.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q2

Using the same subscription parameters but extending the contract to 18 months, recompute the net value per customer and explain exactly why the result changes.

Pricing & MonetizationProduct Analytics & Metrics
Author's notes

Revenue is now 15 paid months at $40, so $600.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, restate the subscription parameters and the original net value calculation to ensure alignment. Then, extend the contract to 18 months, recompute the net value by adjusting for the longer duration, and explain the change by highlighting the time value of money, amortization of upfront costs, and any changes in retention or discounting assumptions.

Pro tip: Always clarify whether the extension changes the discount rate or retention assumptions—often the net value change is driven by the time value of money and cost amortization, not just the extra months. Quantify the impact of each factor to show analytical rigor.

1. Restate the parameters and original calculation

Briefly summarize the given subscription parameters (e.g., monthly price, costs, churn rate, discount rate) and show the original net value per customer for the initial contract length. This ensures everyone is on the same page.

2. Extend the contract and recompute cash flows

Extend the contract to 18 months, keeping all other parameters constant. Recompute the expected cash flows (revenue and costs) for each month, adjusting for any changes in retention or usage patterns if applicable.

3. Apply discounting and sum to get net value

Discount the monthly net cash flows back to present value using the same discount rate. Sum them to obtain the new net value per customer. Compare this to the original net value.

4. Explain the drivers of change

Articulate exactly why the net value changed: (1) additional months of revenue and costs, (2) time value of money (later cash flows are discounted more), (3) amortization of upfront costs over a longer period, and (4) any changes in retention assumptions (e.g., if churn is monthly, the probability of surviving to month 18 is lower).

5. Quantify and conclude

If possible, quantify the contribution of each factor to the overall change. Conclude with the new net value and a clear statement of the primary reason for the difference.

Key Points to Mention

  • Time value of money: discounting future cash flows reduces their present value, and extending the contract adds more distant cash flows that are heavily discounted.
  • Amortization of upfront costs: customer acquisition costs or initial setup costs are spread over a longer period, reducing their per-month impact.
  • Retention/churn assumptions: if churn is modeled monthly, the probability of a customer staying for 18 months is lower, which may reduce expected revenue.
  • Marginal contribution: the additional months may have different marginal costs or revenues (e.g., costs may decrease over time due to learning curves).
  • Discount rate consistency: ensure the same discount rate is used; if the extension changes risk, the discount rate might need adjustment.
  • Sensitivity analysis: discuss how the result changes if key assumptions (e.g., churn, discount rate) vary.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q3

Now extend the contract to 21 months with 10% of customers breaking the contract uniformly across paid months and paying a $100 penalty. Marketing drops to $20/customer variable, overhead becomes $1,000,000/year fixed, installation stays $35. For N customers acquired, derive a formula for the minimum N to break even company-wide.

Pricing & MonetizationProduct Analytics & MetricsProduct Strategy
Author's notes

This is where I slowed down.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, clarify all assumptions and define the revenue and cost components per customer over the 21-month contract, accounting for churn and penalty. Then, set up the break-even equation where total revenue equals total costs, and solve for N. Present the final formula clearly and discuss any simplifying assumptions.

Pro tip: Explicitly state that you assume churn occurs uniformly, meaning each customer has an equal probability of churning in any paid month, and that the penalty is paid at the time of churn. This shows you understand the stochastic nature and can simplify for a deterministic break-even analysis.

1. Define Revenue per Customer

Calculate the expected revenue per customer over 21 months, considering monthly subscription fee, churn rate, and penalty. Let M be the monthly fee. Expected months paid = sum_{k=1}^{21} P(customer pays for month k). Since 10% churn uniformly, the probability of paying for month k is (1 - 0.1*(k-1)/21) for k=1..21? Actually, uniform across paid months means each month has equal chance of being the churn month. So expected number of paid months = (1+2+...+21)/21 = 11.5? Wait, careful: If churn is uniform across paid months, then the churn month is equally likely to be any of the 21 months. So expected months paid = (1+2+...+21)/21 = 11.5. But that would mean on average they pay for 11.5 months. However, if they churn in month k, they pay for k months? Typically, if they churn after paying for month k, they pay for k months. So expected months = 11.5. But also, they pay a penalty of $100 if they churn. The probability of churning is 100% eventually? Actually, 10% of customers break the contract, so 90% complete 21 months. So expected months = 0.9*21 + 0.1*11.5 = 18.9 + 1.15 = 20.05? That seems off. Better: For the 10% who churn, their churn month is uniformly distributed among the 21 months. So expected months for churners = 11.5. For non-churners, 21 months. So overall expected months = 0.9*21 + 0.1*11.5 = 18.9 + 1.15 = 20.05. But wait, if they churn uniformly across paid months, does that mean the churn month is uniform over 1..21? Yes. So expected revenue from subscription = M * 20.05. Plus penalty: 0.1*100 = $10. So total expected revenue per customer = 20.05M + 10.

2. Define Costs per Customer and Fixed Costs

Identify variable costs per customer: marketing $20, installation $35. Also fixed overhead $1,000,000 per year. Since contract is 21 months, fixed overhead for 21 months = 1,000,000 * (21/12) = $1,750,000. Total variable cost per customer = $55. So total cost for N customers = 55N + 1,750,000.

3. Set Up Break-Even Equation

Set total revenue equal to total cost: N*(20.05M + 10) = 55N + 1,750,000. Solve for N: N*(20.05M + 10 - 55) = 1,750,000 => N = 1,750,000 / (20.05M - 45). This is the minimum N to break even.

4. Validate and Interpret

Check that denominator is positive; if 20.05M - 45 <= 0, break-even is impossible. Also, discuss sensitivity to churn assumption and whether penalty is paid at churn or end. Present formula clearly.

Key Points to Mention

  • Expected revenue per customer includes subscription fees over expected paid months plus penalty from churners.
  • Churn is uniform across paid months, so expected months for churners is (1+2+...+21)/21 = 11.5.
  • Overall expected months = 0.9*21 + 0.1*11.5 = 20.05.
  • Variable costs per customer: marketing $20 + installation $35 = $55.
  • Fixed overhead for 21 months = $1,000,000 * (21/12) = $1,750,000.
  • Break-even N = 1,750,000 / (20.05M - 45), assuming M > 45/20.05 ≈ 2.24.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q4

Qualitatively describe the demand vs. price curve and the total profit vs. price curve for this subscription service. Where is revenue maximized versus where profit is maximized, and why can those two points differ?

Pricing & MonetizationProduct Sense & Ideation
Author's notes

Standard micro question but I appreciated it here because it connected to the earlier math.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by defining the demand curve qualitatively—typically downward sloping—and then derive the revenue curve (price × quantity) as an inverted U-shape. Next, introduce costs (fixed and variable) to derive the profit curve, which will also be an inverted U but shifted to the left of the revenue-maximizing point. Explain that the difference arises because maximizing revenue ignores costs, while maximizing profit requires considering marginal cost and marginal revenue.

Pro tip: Emphasize that in subscription services, marginal cost is often near zero, so the profit-maximizing price is typically lower than the revenue-maximizing price to capture more subscribers and maximize total contribution. This shows you understand the business context.

1. Describe the demand curve

Explain that as price increases, the quantity demanded decreases, so the demand curve is downward sloping. Mention that for subscription services, demand may be relatively elastic or inelastic depending on differentiation.

2. Derive the revenue curve

Revenue is price times quantity. As price increases from zero, revenue initially rises because the quantity decrease is proportionally smaller, reaches a maximum, then falls as further price increases cause a larger proportional drop in quantity. Thus, revenue vs. price is an inverted U-shape.

3. Introduce costs and derive profit curve

Profit is revenue minus total costs (fixed + variable). Since costs are positive, the profit curve is also an inverted U-shape but lies below the revenue curve. The profit-maximizing price occurs where marginal revenue equals marginal cost, which is typically at a lower price than the revenue-maximizing price.

4. Compare revenue-maximizing and profit-maximizing points

The revenue-maximizing point is where marginal revenue is zero. The profit-maximizing point is where marginal revenue equals marginal cost. Because marginal cost is positive (even if small), the profit-maximizing quantity is lower and price is higher than at the revenue-maximizing point? Wait, careful: Actually, if marginal cost is positive, profit-maximizing quantity is less than revenue-maximizing quantity, so price is higher. But in subscription services with near-zero marginal cost, the two points may be close. However, the question asks why they can differ: because revenue maximization ignores costs, while profit maximization accounts for them.

5. Explain the difference and business implication

The two points differ because maximizing revenue does not consider the cost of serving additional subscribers. Profit maximization balances the additional revenue from one more subscriber against the additional cost. In subscription services with low marginal costs, the difference may be small, but fixed costs can also shift the profit curve downward without changing the maximizing price.

Key Points to Mention

  • Demand curve is downward sloping: higher price leads to lower quantity demanded.
  • Revenue curve is inverted U-shaped: revenue = price × quantity, maximized where marginal revenue = 0.
  • Profit curve is also inverted U-shaped but lower than revenue curve due to costs.
  • Profit is maximized where marginal revenue = marginal cost, which typically occurs at a lower quantity and higher price than revenue maximization if marginal cost is positive.
  • In subscription services, marginal cost is often near zero, so the profit-maximizing price may be close to the revenue-maximizing price, but fixed costs can affect overall profitability.
  • The difference arises because revenue maximization ignores costs, while profit maximization accounts for both fixed and variable costs.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q5

An insurer sells a 12-month policy at $30/month premium. Servicing costs $3/month. A covered claim pays $8,000 in benefits plus $300 in regulatory costs. There's also a $4/quarter regulatory expense regardless of claims. Let p be the probability of at least one claim in the year. Write and solve the equation where expected profit equals zero.

Pricing & MonetizationProduct Analytics & Metrics
Author's notes

Revenue per policy is $360.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, break down the annual revenue and all costs (both fixed and variable) to express expected profit as a function of p. Then set expected profit to zero and solve for p, ensuring all time units are consistent (monthly vs quarterly).

Pro tip: Always convert all recurring costs to annual amounts before combining them; mixing monthly and quarterly figures is a common mistake that leads to incorrect break-even probabilities.

1. Identify and annualize all revenue and costs

Calculate total annual premium revenue and annual servicing cost. Convert quarterly regulatory expense to annual. Sum fixed costs (those not dependent on claims).

2. Determine claim-related costs

Identify the total cost per claim (benefits + regulatory costs). Recognize that expected claim cost equals p times this total, assuming at most one claim per year (or that p is the probability of at least one claim and the given costs apply to that event).

3. Write the expected profit equation

Expected Profit = Annual Premium Revenue - Annual Fixed Costs - p * (Claim Benefits + Claim Regulatory Costs). Set this equal to zero.

4. Solve for p

Algebraically isolate p. Compute the numerical value, ensuring units are consistent and the result is a probability between 0 and 1.

5. Interpret the result

Explain what the break-even probability means in business terms: the maximum claim frequency the insurer can tolerate before losing money.

Key Points to Mention

  • Annual premium revenue = $30/month * 12 = $360
  • Annual servicing cost = $3/month * 12 = $36
  • Annual regulatory expense = $4/quarter * 4 = $16
  • Total claim cost per claim = $8,000 + $300 = $8,300
  • Expected profit equation: 360 - 36 - 16 - p * 8300 = 0
  • Solving gives p = (360 - 52) / 8300 = 308 / 8300 ≈ 0.0371 or 3.71%

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q6

Given 400 policies observed over a year with 18 claims, compute the MLE for the claim probability, build a 95% Wald confidence interval, explain when the Wald interval breaks down, and suggest a better alternative.

A/B Testing & ExperimentationProduct Analytics & Metrics
Author's notes

MLE is just 18/400 equals 0.045.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by computing the MLE for the claim probability as the sample proportion (18/400 = 0.045). Then construct the 95% Wald confidence interval using the standard formula p_hat ± 1.96 * sqrt(p_hat*(1-p_hat)/n). Finally, discuss the limitations of the Wald interval when p is near 0 or 1 or when n is small, and suggest the Wilson score interval or Clopper-Pearson interval as better alternatives.

Pro tip: Mention that the Wald interval can produce negative lower bounds or exceed 1, which is nonsensical for a probability, and that in practice, especially with rare events, the Wilson interval is more reliable and is often used in A/B testing platforms.

1. Compute the MLE

Calculate the maximum likelihood estimate for the claim probability as the number of claims divided by the number of policies: 18/400 = 0.045.

2. Construct the Wald interval

Use the formula p_hat ± z_{α/2} * sqrt(p_hat*(1-p_hat)/n) with z=1.96 for 95% confidence. Plug in p_hat=0.045 and n=400 to get the interval.

3. Explain when Wald breaks down

Discuss that the Wald interval relies on normal approximation, which is poor when np or n(1-p) is small (e.g., <5 or <10), when p is near 0 or 1, or when sample size is small. Here np=18 and n(1-p)=382, so it's borderline but acceptable; however, for rarer events it fails.

4. Suggest better alternatives

Recommend the Wilson score interval or Clopper-Pearson (exact) interval, which have better coverage properties, especially for small samples or extreme probabilities. Mention that Wilson is often preferred for its balance of performance and simplicity.

Key Points to Mention

  • MLE for a proportion is the sample proportion: p_hat = x/n.
  • Wald interval formula: p_hat ± z * sqrt(p_hat*(1-p_hat)/n).
  • Wald interval can yield bounds outside [0,1] and has poor coverage for small n or extreme p.
  • Rule of thumb: Wald is unreliable when np < 5 or n(1-p) < 5 (some say 10).
  • Wilson score interval is a better alternative, especially for rare events.
  • Clopper-Pearson (exact) interval guarantees coverage but is conservative.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.