I'm glad to see more and more of these types of results. The typical pattern for the past few years at many consumer internet companies is an A/B test, with som

I'm glad to see more and more of these types of results. The typical pattern for the past few years at many consumer internet companies is an A/B test, with some positive short-term metrics and no long-term validation. Any qualitative resistance to the short-term engagement metrics is dismissed, and the change is rolled put. Product managers and software engineering teams parade these metrics wins, get promoted, etc. User experience degrades one a/b test at a time, while "engagement" improves. Glad to see someone is thinking about qualitative user experience and correlating it with long-term metrics. Source: ran an engineering group at a "data driven" consumer internet company that fell victim to these tactics.