Katkat
Back

How should you renew or replace a show?

Capital OneData Scientist · Technical Screen
MediumData ScientistTechnical ScreenAnalytics & Experimentation

Last updated

This question tests the ability to design a decision-making framework for content renewal by balancing historical performance metrics, acquisition costs, and the risk of cannibalization within a streaming portfolio.

The Scenario

You are a Data Scientist for a streaming platform. The leadership team is deciding whether to renew a popular original series for a third season or cancel it to reallocate the budget toward developing new pilot projects.

The Data

You have access to the following metrics:

  • subscriber_acquisition_count: How many new users joined the platform specifically because of the show.
  • churn_rate_delta: The difference in churn rates between users who watched the show versus those who did not.
  • production_cost: The total cost to produce a new season.
  • clv_estimate: The projected lifetime value of an average subscriber.
  • audience_overlap: The percentage of the show's audience that also watches your platform's other top-performing shows.

Your Task

  1. Define a mathematical framework to determine the Return on Investment (ROI) for renewing this show. How do you account for both direct acquisition and the mitigation of churn?
  2. Explain how you would account for 'cannibalization'—if users are already subscribed to watch other content, how do you attribute value to this specific show?
  3. Propose an A/B test or a quasi-experimental design (such as a Synthetic Control Method) to validate whether the show's presence actually causes a reduction in churn, rather than just correlating with low-churn users.

Constraints

  • Focus on the decision-making logic and the statistical methodology rather than raw code.
  • Consider the risk of 'sunk cost fallacy' in your analysis.

Hints

Hint 1

Consider calculating the 'marginal subscriber contribution' by comparing the CLV of acquired users against the production cost per subscriber.

Hint 2

Think about using the audience overlap metric to penalize the show's value if viewers would likely watch other high-performing shows on the platform anyway.

Hint 3

For the causal aspect, consider how you might define a control group if you are forced to make a binary keep/drop decision for the entire platform.

Follow-up questions

  • How would your model change if the goal shifted from 'subscriber growth' to 'long-term brand prestige and award-winning potential'?
  • If the show has a high audience overlap, how does that change the risk profile of cancelling it versus a show with a unique, niche audience?
  • How would you communicate the trade-off between the high-certainty returns of a proven show versus the high-variance potential of a new pilot to non-technical stakeholders?