Data and Business Outcomes Questions

This topic focuses on converting data analysis and insights into actionable business decisions and measurable outcomes. Candidates should demonstrate the ability to translate trends into business implications, choose appropriate key performance indicators, design and interpret experiments, perform cohort or funnel analysis, reason about causality and data quality, and build dashboards or reports that inform stakeholders. Emphasis should be on storytelling with data, framing recommendations in terms of business levers such as revenue, retention, acquisition cost, and operational efficiency, and explaining instrumentation and measurement approaches that make impact measurable.

HardTechnical

54 practiced

You need to estimate the causal effect of a push-notification campaign on daily active users (DAU) but notifications were targeted by a rule that used past activity. Explain how regression discontinuity or instrumental variables might be used here and the assumptions required.

EasyTechnical

42 practiced

Explain what a cohort analysis is. Give one concrete example of a cohort definition for a subscription product and describe the business question that cohort analysis would answer.

MediumTechnical

53 practiced

Explain how uplift modeling differs from standard conversion modeling. Give one business scenario where uplift modeling is preferable and describe the data and experiment design required.

HardTechnical

43 practiced

Write pseudocode or Python-like code to compute weekly cohort LTV (sum of revenue per cohort member over N weeks) from two tables: users(user_id, acquired_date) and purchases(user_id, purchase_date, amount). Optimize your approach for millions of users and purchases.

Sample Answer

Approach: compute cohort by week of acquired_date and sum each user's purchases within N weeks post-acquisition; use scalable joins and aggregation (windowed map-reduce or SQL on distributed engine). Pseudocode (Python-like, using chunked processing and groupby to scale):

python

# assume purchases and users are iterators over parquet files
from datetime import timedelta
N_weeks = 12
cohort_revenue = defaultdict(lambda: [0]*N_weeks)
# build user cohort map in memory (user_id -> cohort_week)
for user in users_stream():
    cohort = user.acquired_date.isocalendar()[:2]  # (year, week)
    user_cohort[user.user_id] = cohort
# stream purchases (memory-light)
for p in purchases_stream():
    cohort = user_cohort.get(p.user_id)
    if not cohort: continue
    week_offset = (p.purchase_date.date() - cohort_start_date(cohort)).days // 7
    if 0 <= week_offset < N_weeks:
        cohort_revenue[cohort][week_offset] += p.amount
# after streaming, divide by cohort sizes (from users) to get per-member sums
for cohort, sums in cohort_revenue.items():
    size = cohort_size[cohort]
    ltv = [s/size for s in sums]
    emit(cohort, ltv)

Key optimizations: store user->cohort map partitioned, stream purchases, use distributed processing (Spark SQL: partition by cohort, aggregate), pre-aggregate by partition to reduce shuffle, and persist intermediate results.

MediumTechnical

48 practiced

A retention curve shows that users acquired via Channel A have higher short-term retention but lower 6-month retention than users from Channel B. Propose hypotheses for why this might occur and design analyses to test each hypothesis.

Unlock Full Question Bank

Get access to hundreds of Data and Business Outcomes interview questions and detailed answers.

Join thousands of developers preparing for their dream job.