Behavior is different on Pinterest: people use it to plan, which differentiates it from other platforms. They’re in discovery mode, in feed and in search. 75% of Pinterest users say it inspires them to try new brands/products. Pinterest partnered with Stanford and surveyed 6K respondents to determine how activity on Pinterest related to real world intent. Two key topics for intent prediction: 1) Temporal range; 2) Goal specificity. Findings: The number of searches and the mean # of keywords in the search, the higher the more likely there was specific goal within a short timeframe. Intent can be modeled within 2 minutes of user activity on the platform. Current activity within a user’s session is the most predictive feature of intent, more predictive than demographics or historical activity. [Sponsored]