What happened
Meta launched "Dear Algo" on Threads, a natural language interface for feed customisation. Users submit text prompts to temporarily adjust recommendation weights for specific topics, keywords, or creators. The system modifies algorithmic ranking of the "For You" feed based on explicit instructions. This feature supplements passive signals like likes and shares. It allows users to request more or less of specific content categories for defined periods.
Why it matters
Product managers and ad buyers face shifting engagement patterns because users now explicitly override passive algorithmic signals. This shift from implicit to explicit preference collection reduces reliance on historical click data, therefore shortening the feedback loop for content relevance. This follows TikTok’s November 2025 content control launch and Firefox’s February 2026 AI controls, marking a broader industry move toward user-steered AI. Result: platform engineers must now reconcile conflicting signals between long-term engagement data and short-term prompts.
