← Meta Interview Insights

Meta·Data Scientist·Technical Phone Screen·Intermediate

Intermediate
May 2026

Summary

Probability and stats question for a DS role at Meta, focused on ad insertion strategies. Pretty clean interview, nothing too wild, but the math needs to be tight or you'll fumble the comparison part.

Questions Asked (3)

Q1

Under Strategy A, each post independently has a 4% chance of being an ad. What is the expected number of ads shown in 100 posts?

Product Analytics & Metrics
Author's notes

Straightforward linearity of expectation.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Recognize that the expected number of ads is the sum of independent Bernoulli trials, so multiply the number of posts (100) by the probability of an ad (0.04). State the formula E[X] = n * p and compute 100 * 0.04 = 4. Then briefly explain why linearity of expectation applies even without independence, but here independence makes it straightforward.

Pro tip: Don't just give the number—connect it to a business insight, like how this expected value informs ad load or user experience metrics, and mention that variance (n*p*(1-p) = 3.84) gives a sense of variability around the mean.

1. Identify the distribution

Recognize that each post is a Bernoulli trial with success probability p = 0.04, and the total number of ads follows a Binomial(n=100, p=0.04) distribution.

2. State the expectation formula

For a Binomial distribution, the expected value is E[X] = n * p. Alternatively, use linearity of expectation: sum the expected values of each post.

3. Compute the expected value

Plug in n = 100 and p = 0.04 to get E[X] = 100 * 0.04 = 4 ads.

4. Interpret and extend

Explain that on average, 4 ads will be shown. Optionally, mention the variance (3.84) or standard deviation (~1.96) to describe typical fluctuations.

Key Points to Mention

  • Binomial distribution and its parameters (n, p)
  • Linearity of expectation and why it applies
  • Independence assumption and its role
  • Calculation: 100 * 0.04 = 4
  • Variance and standard deviation for context
  • Business implication: expected ad load and potential user impact

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

Q2

Strategy B guarantees exactly one ad per block of 25 posts. How does the expected number of ads per 100 posts compare between Strategy A and Strategy B?

Product Analytics & MetricsA/B Testing & Experimentation
Author's notes

Both give you 4 expected ads per 100 posts, which feels like a trick until you actually work it out.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, clarify the distributional assumptions for Strategy A (e.g., independent posts with probability p of being an ad) and compute its expected ads per 100 posts as 100p. Then, for Strategy B, recognize that the guarantee of exactly one ad per 25-post block means the expected number of ads per 100 posts is exactly 4, regardless of any randomness. Finally, compare the two expectations and discuss implications for variance and practical significance.

Pro tip: Highlight that Strategy B eliminates variance in ad count per block, which can be beneficial for user experience consistency, but may reduce overall expected ads if p > 0.04. Always tie the comparison back to business metrics like revenue or user engagement.

1. Clarify Strategy A's ad model

Assume each post independently has probability p of being an ad. Then the number of ads in 100 posts follows Binomial(100, p), so the expected number of ads is 100p.

2. Compute Strategy B's expected ads

Strategy B guarantees exactly one ad per 25 posts. Thus, in 100 posts (4 blocks), there are exactly 4 ads, so the expected number is 4.

3. Compare expectations

Compare 100p to 4. Strategy A has higher expected ads if p > 0.04, lower if p < 0.04, and equal if p = 0.04.

4. Discuss variance and practical implications

Strategy B has zero variance in ad count per block, while Strategy A's variance is 100p(1-p). This affects predictability and user experience. Consider business impact: revenue, engagement, and ad load.

Key Points to Mention

  • Expected value calculation for binomial distribution: E[X] = np for Strategy A.
  • Strategy B's deterministic placement yields exactly 4 ads per 100 posts.
  • Threshold p = 0.04 where both strategies have equal expected ads.
  • Variance comparison: Strategy B has no variance, Strategy A has variance np(1-p).
  • Implications for A/B testing: need to consider both mean and variance when choosing metrics.
  • Business context: ad load, user experience, and revenue trade-offs.

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

Q3

Still under Strategy A, what is the probability that a user sees at least one ad in the next 25 posts?

Product Analytics & Metrics
Author's notes

Classic complement trick.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, clarify the assumptions about Strategy A, such as the probability of an ad appearing in a single post and whether ad placements are independent. Then, model the number of ads in 25 posts using a binomial distribution and compute the probability of at least one ad as 1 minus the probability of zero ads.

Pro tip: Always state your assumptions explicitly and consider edge cases like p=0 or p=1; this shows rigor and prevents miscommunication. Also, relate the result to business impact, such as user experience or ad revenue.

1. Clarify assumptions

Confirm the probability of an ad in a single post under Strategy A and whether ad events are independent across posts. If not given, state reasonable assumptions.

2. Define the random variable

Let X be the number of ads in 25 posts. Under independence and constant probability p, X follows a Binomial(n=25, p) distribution.

3. Compute probability of at least one ad

Use the complement rule: P(X ≥ 1) = 1 - P(X = 0) = 1 - (1-p)^25.

4. Plug in values and interpret

Substitute the given p (or a range) to get a numerical answer. Discuss how the probability changes with different p and what it means for user ad exposure.

Key Points to Mention

  • Independence assumption for ad placements across posts
  • Binomial distribution as the model for number of ads
  • Complement rule to simplify 'at least one' probability
  • Sensitivity of the result to the value of p
  • Business implications: user experience, ad load, and revenue
  • Edge cases: p=0 (no ads) and p=1 (ads in every post)

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