Notion's long road to product-market fit: lessons from Ivan Zhao

Executive overview

Most tools lock users into predefined workflows. Notion was built on the opposite premise: give people Lego bricks and let them build whatever they need.

Getting there took four lost years, multiple full rebuilds, and a near-death database crisis during COVID. The core tension Ivan navigates throughout — between building for personal values and building for what the market actually wants — is what drove both the failures and the eventual success.

The hardest problem in building a horizontal product is not the product itself — it's knowing when to show the brick and when to sell the box.

The lost years: 2013–2017

• First version was a developer tool built on the belief that anyone could create software — almost nobody cared
• Took two years to accept the realization: most people wake up with a report due; they don't want to build tools
• Pivot insight: hide the vision (programmable software) inside the form factor people already use (productivity tools)
• "People don't want to eat the broccoli. Give them the sugar. Hide the broccoli inside."
• Built on Google's Web Components — technology turned out too unstable; had to scrap the codebase and restart
• Laid off the whole team, back to just Ivan and co-founder Simon; moved to Japan to reset and rebuild
• Survived financially through staying lean, subleasing the SF office, and a loan from Ivan's mother
• Product-market fit arrived gradually — no single moment, just a slow ramp of people caring and paying

Staying lean as a deliberate strategy

• Profitable for several years; most of the raised capital remains unspent
• First salesperson hired after $10M ARR; first PM hired around 50 employees
• Internal metric: talent density (revenue per employee), not headcount
• Small teams reduce alignment overhead — communication cost scales faster than output
• Notion itself is the internal operating system; using the product to run the company keeps headcount low
• Mental model: "small bus" — easier to turn corners, accelerate, and maneuver than a large one
• Leader's job is choosing who sits on the bus; seat selection determines speed and culture

Craft, trade-offs, and office design

• Company philosophy: craft (skill + taste applied to trade-offs) and values (what you want the world to have)
• No free lunch in product: every decision trades something; identifying what to trade is the craft
• Built to last — conference rooms named after timeless tools: original Macintosh, Lamy 2000 pen, Toshiba rice cooker
• Office designed to feel like an artist's studio or home — warm walls, home furniture, no harsh overhead lighting
• Shipping "non-Lego" features (e.g., hard-coded sprint concept) caused internal and customer friction; took 18 months to course-correct
• Rule of thumb: if a feature fits the Lego system, the codebase works with you; if it doesn't, it works against you

The COVID database crisis

• For years, Notion ran on a single Postgres instance — Simon's principle of avoiding premature optimization
• During COVID, usage surged and the largest available Postgres instance was nearly full
• Doomsday clock: weeks to months before the database would run out of space and Notion would shut down entirely
• All engineering halted; every engineer redirected to sharding the database
• Resolved just in time — the lesson: don't optimize prematurely, but plan ahead enough to avoid existential deadlines

Building horizontal: the joy and the suffering

• Horizontal products take longer to find fit but face no ceiling — the entire software market is addressable
• Segmentation is critical: most users want a Lego box (ready-made set), not loose Lego bricks
• Enterprise customers need "solutions" — named use cases with clear value props, not primitives
• Stayed too long in brick-seller mode; shifting to solution-seller mindset was a hard-won lesson
• Growth strategy: B2C2B — consumers adopt Notion for notes/docs, then bring it to work
• Notes and docs gave a 1B+ person top-of-funnel; calendar and email extend the same logic
• Horizontal + AI is a powerful combination: unified data in one place makes AI reasoning dramatically better
• Three-phase AI roadmap: (1) AI writing inside Notion, (2) AI Q&A across all Notion data, (3) AI agents that assemble Lego bricks into custom workflows on demand

Leadership and building for values

• Direct communication style — closer to East Coast than West Coast; values bluntness over false positivity
• Scaling from one-to-one to one-to-many storytelling was an unnatural but necessary craft to develop
• Treats each new company stage as a new game requiring new skills — "infinite games are more fun"
• Building for personal values is the more durable energy source across dark years; building purely for business produces commodities
• Tools are extensions of humans — and once built, they reshape the people who use them
• Goal: amplify creativity and beauty in software, not just latch onto base human impulses

Advice for founders

• Reset without fear: better abstractions compound faster than accumulated code; you recover lost ground faster than expected
• Momentum is dangerous — it carries you forward even after a wrong turn; stop and rethink deliberately
• Balance personal vision with market reality: too much self = research project; too much market = commodity
• Sleep is a legitimate reset mechanism; low lows and low highs are a structural advantage for long-haul founders
• Look outside tech for patterns: history, other industries, and old books contain trade-offs and shapes that map directly onto product and business problems
• Read in the complex systems domain — emergent behavior from few primitives is the mental model behind Notion's entire architecture