The rise of the connected consumer has led many to pronounce the classic purchase funnel and its attitudinal metrics dead, replaced by online behavior metrics. Although online behavior metrics offer the benefits of timeliness and passive tracking, classic attitude survey metrics use  representative samples and have improved over decades of market research. Which metrics are best at explaining and predicting sales? We address this question for 36 brands over 15 categories, including services, durables, and fast-moving consumer goods. Dynamic system models capture interactions among metrics, between marketing and metrics and between metrics and sales. We find that both types of metrics matter, but that online behavior metrics excel in sales explanation, while attitude survey metrics excel in sales prediction. Moreover, online action does not simply follow from attitudes, it also drives them. Therefore, we propose an integrative model of consumer actions and attitudes as a boulevard of fast action enclosed by slower evolving attitudes. Managers can use both online and offline marketing to drive either lane in this boulevard:  while online advertising is most effective in lifting many attitude survey and online behavior metrics, TV ads drive engagement metrics of social media and page views.