Match a job Paths Subjects Questions Quizzes Pricing
Advanced Open Pro

Diagnose a Stale Surrogate Caused by Insufficient Exploration

Six months after launch, an analyst notices that a specific content category ("long-form explainers") that scored well on the surrogate model at launch has steadily lost feed placement, even though nothing about the category's actual quality seems to have changed.

  1. Propose a hypothesis rooted in the ranker/surrogate feedback loop that would explain this pattern.
  2. What data would you pull to confirm or reject that hypothesis?
  3. What's the fix, and why does it need to happen at more than one funnel stage?

Share this question

← Back to Case Study: Long-Term Engagement Recommender practice

We use cookies for product analytics to improve OmniAtlas. See our Privacy Policy.