Ivan Zhao on Notion and Building Horizontal Products

Source: Lenny’s Podcast Speaker: Ivan Zhao Source URL: https://www.lennysnewsletter.com/p/ivan-zhao

Key ideas

Sugar-coated broccoli: Notion’s breakthrough was hiding its real vision (everyone can create software) inside a productivity tool people already wanted. Vision ≠ product form factor.
The lost years: Notion’s first 3–4 years were iterating on the wrong form factor and then the wrong technical foundation. Staying lean and conviction in the vision (not the specific product shape) is what allowed survival.
Talent density over headcount: no salespeople until $10M ARR, no PM until 50 people. The goal is revenue-per-employee density; systems and tools > brute-force hiring.
Lego bricks vs. Lego boxes: horizontal tools must sell solutions (boxes) to enterprises while maintaining primitives (bricks) underneath. Hardcoding solution-specific features (e.g., sprints) violates the primitives model and causes “organ rejection.”
Craft + value as a durable energy source: building something authentic to your own values and aesthetics is more sustainable than optimising for business outcomes. The balance: too much self = research project; too much business = commodity.

Origin: from Urumqi to Notion

Ivan grew up in Urumqi (northwest China, 4M people). Moved to Beijing; won second place in the Beijing Information Olympiad (programming contest), which secured admission to a top school. Later moved to Canada at 16; learned English via SpongeBob and The Simpsons. Studied art and science, not computer science (already knew how to code).

The founding insight: making websites for artist friends who couldn’t code led to reading Douglas Engelbart’s “Augmenting Human Intellect.” The 1960s–70s computing pioneers (Engelbart, Alan Kay, Ted Nelson) believed there should be no separation between builders and users of software. Steve Jobs saw the Xerox Alto’s GUI but missed its deeper power: the Smalltalk object system in which everything was malleable. The PC era chose the application framework over the object framework; every SaaS tool today is a downstream consequence of that choice.

Notion’s mission: Lego for software — reverse the App Framework decision; give non-programmers the ability to build software for their daily work.

The lost years (2013–2017)

Version 1: easy developer tool — let anyone create software. Result: most people don’t care. “The world wakes up, has a report due, needs to get their job done.”

Lesson 1 — sugar-coated broccoli: hide the vision inside a product people already understand and use. What productivity tools do people use every day? Documents. That became Notion. The broccoli (everyone creates software) is inside the sugar (productivity tool).

Version 2: rebuild on the wrong technical foundation (Web Components, a losing competitor to React). Bugs were untraceable. Solution: lay off everyone, restart the codebase in Tokyo with two people (Ivan + Simon), building 18 hours a day. “Some of the happiest moments.”

On resetting: progress through better abstractions compounds faster than you’d expect. The Smalltalk kernel was ~100 lines of code. Small, right abstractions > large, wrong ones. “You can create progress through better abstractions and that thing compounds faster, can catch up to all the things you built much quicker than you ever thought.”

PMF signal: not a binary moment. Gradual: “People who care about this now. Revenue. Investors starting to knock on doors.” Investor (Shana Fisher) interrupted a meeting series to say: “Ivan, what are you doing? You clearly don’t need money. You’re just trying to feel good from external validation.” They went back to building-only mode.

Staying lean

No salespeople until >$10M ARR.
No dedicated PM until 50 people.
Team still in the hundreds while competitors have thousands.
Notion runs almost entirely on Notion internally.
Metric: talent density (revenue per employee), not headcount.

Small bus metaphor: a smaller bus turns corners faster, accelerates faster. The people around you on the bus determine the speed of the whole bus. As a leader, you choose who sits next to you.

On overhead: the real cost is internal communication — getting people’s minds aligned on things. Systems and tools can often replace people; systems compound.

Craft and values

Ivan’s company philosophy: craft + values. Craft = applied taste + technical know-how to make clever trade-offs. Values = what you want the world to have more of.

The balance: building entirely for your own values = research project (no users). Building entirely for the market = commodity. Notion found the middle by hiding the vision (broccoli) inside what the market understands (sugar).

Organ rejection: Notion shipped sprints as a hardcoded product feature (rather than as a Lego primitive) to compete in project management. Result: underpowered relative to specialised tools; code didn’t fit the rest of Notion’s architecture; felt wrong to engineers and users. Took 9–12 months to reverse. Lesson: if you build Lego-way inside the codebase, the system works for you. If you build non-Lego-way, the system works against you.

Horizontal products

What makes horizontals hard: the product must solve multiple use cases without becoming a different specialised tool for each. Most horizontal companies take 8–10 years to find meaningful traction.

Notion’s solution — B2C2B: start with a use case with 1B+ users (document notes). Personal users discover progressively more powerful Notion capabilities and bring it to work. ~50% of B2B customers were prior personal users. The strategy: find a horizontal use case with massive top-of-funnel, then layer in B2B solutions on top.

Lego bricks vs. Lego boxes:

Hardcore users want bricks (primitives, composability).
Most users — especially enterprise — want boxes (ready-made solutions).
Horizontal builders must hold both in their head: sell boxes to go-to-market; build bricks to avoid pigeonholing. Compromising the bricks (sprints case) costs architectural integrity and competitive differentiation.

Bundling/unbundling pattern (Romance of the Three Kingdoms): “Empires long united must divide, long divided must unite.” Software has cycled: pre-PC (paper) → PC era (fragmented apps) → Microsoft suite (bundled) → SaaS (unbundled, now ~100 tools per company) → AI era (rebundling). Horizontal + AI-native = structural tailwind for Notion’s next phase.

AI: three phases:

AI writer (writing inside documents — easy first slice given existing surface area).
AI Q&A / connectors (reason across everything in Notion + external integrations).
AI agents on top of Notion’s Lego blocks — the most exciting: “We spent 5–6 years building all those Lego blocks for knowledge work. If we put an AI coding agent on top, you can create any kind of knowledge software for whatever vertical use case you need.”

Tools and human potential

Favourite quote (Marshall McLuhan): “We shape our tools. Then after, our tools shape us.” Conference rooms named after timeless tools: original Macintosh, Lamy 2000 pen, Toshiba rice cooker, Sony transistor radio. The question: what part of human nature are you amplifying? Notion’s answer: creativity and beauty.