The cold-start problem is the recommender's blind spot at the edges: a brand-new user has no viewing history to learn from, and a brand-new title has no audience signal yet, so behavioral methods like collaborative filtering have little to work with.
Platforms mitigate it from both sides. For new users: onboarding taste prompts, popular and trending defaults, and fast learning from the first few plays. For new titles: content-based filtering from metadata, editorial placement, and similarity to known titles. Handling cold start well matters because the first session shapes a new subscriber's impression - and the first weeks are when churn risk is highest.
