The Contrarian Playbook: How Superhuman Found PMF by Doing Less, Not More
Rahul Vohra, CEO of Superhuman, spent years building a product that broke almost every conventional PM rule — and the metrics proved him right.
Filter your feedback ruthlessly
Most teams treat customer feedback as a democracy. Vohra argues this is exactly wrong. His PMF measurement starts with one question: How would you feel if you could no longer use this product? Only the 'very disappointed' cohort tells you anything actionable. A customer who'd shrug at losing your product is not a signal — they're noise. Every piece of feedback should be filtered through whether it comes from someone who'd be devastated to lose you. Everything else, no matter how loudly it's expressed, is a distraction that produces a sloppy product that tries to please everyone and delights no one.
Stay manual longer than feels comfortable
When Superhuman was growing, Vohra insisted on one-to-one concierge onboarding for every single new user — at peak, 20 people were doing this full-time. The results were exceptional across every metric: engagement, retention, NPS, virality. The mechanism isn't mysterious: manual onboarding creates super-fans, and super-fans create real word-of-mouth. Viral mechanics don't build brands. Super-fans do.
Know when to stop — and why it's harder than it looks
The signal to stop manual onboarding isn't ARR or headcount — it's when a distinct personality type who refuses one-on-one sessions becomes a material segment you can't afford to lose. The transition to self-service felt nearly impossible at Superhuman because the entire company DNA was built around live, total information transfer. What made manual onboarding work is precisely what makes it hardest to replicate in software.
Name the slowdown accurately before you fix it
When Superhuman noticed a perceived slowdown, Vohra's diagnosis was structural: most product work can be classified as either solution deepening (making the product better for people you already serve) or market widening (making it work for people you don't yet serve). The dangerous failure mode is when an org silently shifts entirely to solution deepening. It feels productive. It generates beautiful roadmap slides. But it doesn't move ARR. And it's the politically safer argument in every planning meeting, which is exactly why it's a trap.
Own one attribute completely
Superhuman didn't compete on features — it competed on one attribute: speed. Single-attribute positioning focuses every product decision, makes the brand legible, and ensures you're building something the 'very disappointed' segment can evangelize with a one-sentence pitch.