| url | https://review.firstround.com/how-superhuman-built-an-engine-to-find-product-market-fit |
|---|---|
| raw | raw/superhuman-pmf-engine.txt |
Rahul Vohra’s essay on turning product-market-fit from an abstract feeling into a measurable, optimizable metric. The most actionable PMF methodology published to date.
The problem
Andreessen’s definition of PMF (only-thing-that-matters) is vivid but describes lagging indicators — by the time investment bankers are staking out your house, you already have PMF. Vohra needed a leading indicator to know where Superhuman stood before launch, and a systematic process to improve it.
The Sean Ellis test
Sean Ellis (early growth at Dropbox, LogMeIn, Eventbrite) found a leading indicator: ask users “How would you feel if you could no longer use the product?” and measure the percent who answer “very disappointed.”
The magic number is 40%. Companies below 40% almost always struggle to grow. Companies above it almost always have strong traction. When Hiten Shah surveyed 731 Slack users in 2015, 51% said “very disappointed” — confirming Slack’s PMF when it already had ~500K paying users.
The four-step engine
Superhuman started at 22% very disappointed. The engine:
1. Segment to find supporters. Group survey responses by the first question. Look at personas in the “very disappointed” segment to narrow the market. Superhuman’s score jumped from 22% to 33% just by focusing on the right users (founders, managers, executives, BD). Then build a High-Expectation Customer (HXC) profile using Julie Supan’s framework — the most discerning person within your target demographic.
2. Analyze feedback to convert fence-sitters. Why do the “very disappointed” users love the product? (Speed, focus, keyboard shortcuts.) What holds the “somewhat disappointed” users back? Key move: ignore the “not disappointed” users entirely — they’ll request distracting features and churn anyway. Focus on “somewhat disappointed” users whose main benefit aligns with your core value proposition. For Superhuman: mobile app, integrations, calendar.
3. Build a split roadmap. Half the roadmap doubles down on what lovers love (more speed, more shortcuts, more automation). The other half addresses what holds fence-sitters back (mobile, integrations, calendar). Neither alone is sufficient — only doubling down won’t increase the score; only addressing gaps lets competitors overtake you.
4. Track and repeat. Make the “very disappointed” percentage the team’s most important metric. Survey new users continuously (never re-survey). Within three quarters, Superhuman went from 22% to 58%.
Key claims
- PMF is measurable via the Sean Ellis 40% test — a leading indicator, not a lagging one.
- Segmentation before optimization. Your aggregate score hides pockets of strong fit. Narrow the market first.
- Politely disregard users who wouldn’t be disappointed without your product. Their feedback leads you astray.
- Split your roadmap 50/50 between deepening love and removing barriers.
- As you grow, PMF score may drop as you encounter more demanding users. Network-effect businesses get natural defense; SaaS companies must keep improving the product.
Connections
Builds directly on Andreessen’s only-thing-that-matters and operationalizes it. The “narrow the market” advice echoes Paul Graham’s “make something a small number of people want a large amount” (hackers-and-painters, graham-on-power). The segmentation approach connects to Moore’s beachhead strategy in crossing-the-chasm-concept. The emphasis on retention over growth aligns with pmf-roadmap and duolingo-growth. The HXC framework parallels Chen’s atomic-network — find the smallest viable user group that loves you.