AI-generated personas are fast and free, and the research comparing them to real participants keeps finding the same flaws: less variance, idealized behavior, flattened segments and a strong urge to...
Recent Posts
Network effects get credit for most of the value in tech, yet only about a third of billion-dollar tech companies have one at their core. Three tests separate a real network effect from a decorative...
About one candidate in three is cut after reference checks, and for senior roles the call that matters is the backchannel you never listed. Product managers leave more of those witnesses than almost...
Shopify deleted 12,000 recurring meetings in one sweep, and meeting time stayed 14% lower months later. Product managers create more standing syncs than anyone, so they should be the ones setting end...
Knight Capital lost $460 million in 45 minutes because of a feature flag it stopped using in 2003. Why flag debt is a product problem, and how to put feature flag cleanup into your definition of done.
Most Product Experiments Fail. Some of the Winners Are Fake Too.
At Airbnb, Booking.com and Bing, only 8 to 15 percent of experiments win. Here is what that failure rate, and the false positives hiding among the winners, means for product discovery.