Entrepreneurship KB

Powered by Agentic Workers

Case Study: Linear — $1.25B on $35K of Marketing

9 min read

Case Study: Linear — $1.25B on $35K of Marketing

Linear is the case study for founders who think “growth” must be bought. Founded in January 2019 by three Finns frustrated with Jira, Linear reached a $1.25B valuation (June 2025) on ~$35,000 of lifetime paid marketing spend — roughly 0.5% of revenue. They were profitable by 2021 and have negative lifetime burn: more cash in the bank than they’ve ever raised. They defied nearly every SaaS growth orthodoxy and won on craft, focus, and opinionated design.

Timeline

YearEventScale
2014Karri Saarinen & Jori Lallo’s Kippt (YC S12) acquired by CoinbaseFounder team forms
2015-2018Saarinen at Airbnb (Principal Designer, DLS lead); Artman at Uber; Lallo at CoinbaseDay jobs
Jan 2019Linear founded after bar meeting of three frustrated engineers/designers3 founders
Apr 2019Private beta launches with waitlist; ~10 users invited per week~10K waitlist
Nov 2019$4.2M Seed led by Sequoia (discovered Linear via Twitter)~12 months old
Jun 2020Public launch after 14 months of private beta~1,000 paying customers
2021ProfitableNegative lifetime burn
Dec 2020$13M Series A led by Sequoia-
Sept 2023$35M Series B led by Accel, ~$400M valuation~$100M ARR era
Jun 2025$82M Series C led by Accel, $1.25B valuation15,000+ companies, ~178 employees

Total raised: $134.2M. Profit +280% YoY in the most recent reported year.

The Founders

  • Karri Saarinen (CEO) — Principal Designer at Airbnb, led the Design Language System. Previously founding designer at Coinbase.
  • Tuomas Artman (CTO) — Staff engineer at Uber on mobile platform. Built Linear’s custom sync engine.
  • Jori Lallo (CPO) — Founding engineer at Coinbase (post-Kippt acquisition).

All three met in the Helsinki startup scene before the Bay Area. Saarinen and Lallo had already co-founded and exited Kippt (YC S12 → Coinbase). The founding moment: at a bar in early 2018, the three realized they all hated Jira. Saarinen had literally built a Chrome extension at Airbnb to apply custom CSS that made Jira less painful — about 100 Airbnb employees used it. Before quitting, they spent roughly a year interviewing colleagues about tool pain.

Saarinen: “We use all these software development tools at our companies and all of them are quite bad.”

Mapping to Frameworks

distribution: The $35K Growth Engine

Linear is the canonical counter-example to the “you need a marketing team” thesis. From founding through Series C (2019-2025):

  • $35,000 lifetime paid marketing spend (~0.5% of revenue)
  • 3 years before hiring their first marketer
  • 4-person sales team at Series C (~20% of headcount is usual; Linear is ~2%)
  • Zero A/B tests, ever
  • No growth hacking — no referral bribes, no viral loops beyond team-to-team expansion

What replaced paid marketing:

  1. Design-as-distribution — every screenshot looked good enough to share. The changelog itself was a distribution channel. Even error states were designed to be screenshot-worthy.
  2. Twitter as public changelog — @linear and @karrisaarinen tweeted build-in-public style updates. The Sequoia partner who led the seed round discovered Linear entirely through her Twitter network’s enthusiasm.
  3. The Linear Method as ideology — publishing their philosophy turned adoption from feature-comparison into belief-adoption. Teams didn’t pick Linear because it had more features; they picked it because they agreed with how Linear thinks.
  4. Team-to-team viral loop — natural org expansion replaced growth hacks. Once one team on Linear, adjacent teams followed.
  5. Community — 16,000+ member Slack community.

do-things-that-dont-scale: Handpicked Beta Cohorts

For 14 months (April 2019 → June 2020), Linear was invite-only. Saarinen personally handpicked roughly 10 users per week from the waitlist, reading survey answers to match cohort fit. The loop:

  1. Invite 10 users
  2. Fix their complaints
  3. Invite the next 10

By the time Linear opened to the public, they had ~1,000 paying customers — every single one converted manually, one invite at a time. Early beta users included founders who would later run Cohere, Runway, and Ramp. This is Graham’s thesis executed at professional scale.

focus: Opinionated Product as Weapon

Linear’s list of things they refuse to do is the article. They have no:

  • Custom fields per project
  • Custom workflow states (backlog → todo → in progress → done is fixed)
  • Custom issue types
  • Time tracking
  • Deep nested sub-projects
  • Jira-style permission schemes
  • Per-user dashboards with drag-drop widgets

Saarinen: “I don’t think you can build the optimal tool if it’s very flexible or endlessly customizable. Design something for someone. It’s hard — impossible even — to design something really good for everyone.”

This is Altman’s focus and Rabois’ editing made product. Every “no” is a positioning decision. Linear optimizes for the engineer using the tool; Jira optimizes for the admin configuring it.

moats: Craft, Speed, and Taste

Linear’s moats don’t appear on any standard list:

  • Craft as a moat — Saarinen: “We started with quality… people actually noticed, because it’s a rare approach — especially for startups.” When the competitive field is “slow, ugly, bloated tools,” shipping beautiful, fast software becomes a durable differentiator. Competitors can copy features; they cannot copy taste without the team that has it.
  • Performance as a moat — Every interaction targets sub-100ms. Linear’s API issue updates clock ~47ms vs Jira’s ~3.2 seconds — a 68x difference. Artman’s custom sync engine (IndexedDB local-first, optimistic mutations, GraphQL writes, WebSocket realtime sync) makes this architectural, not just optimization. Competitors cannot bolt this on.
  • Opinionated defaults as a moat — because Linear makes all the decisions, switching to Linear is a simplification. Switching away is a regression. Customization-heavy incumbents can’t easily copy this — their customer base depends on their flexibility.

Quote: “Quality is a choice we can make every day.”

product-led-growth: PLG That Charged From Day One

Linear is a PLG company, but with contrarian twists on the standard playbook:

  • Charged from day one — no generous freemium. Free tier capped at 250 active issues (you’ll hit it in a week as a serious team).
  • Bottom-up adoption — engineers discover Linear individually, bring it to their team, teams bring it to the company. OpenAI, Vercel, Cursor, Ramp, Perplexity, Cohere, Runway, Retool, Scale AI, Cash App all adopted this way.
  • Time-to-aha under 10 minutes — the product sells itself within the first session.
  • 80%+ DAU rate — retention that looks more like a consumer app than a B2B tool.

hiring: Paid Work Trials, Not Whiteboard Interviews

Linear hires through paid 1-week work trials in the real Linear codebase. This evaluates product sense and craft, not algorithm trivia. They hire product engineers — generalists with strong product intuition — and have only ~2 dedicated PMs at 1B+ scale. Engineers own product decisions; engineers rotate through customer support directly.

Team composition:

  • First hire: 6 months post-launch (engineer from Coinbase network)
  • ~30 employees at Series A
  • ~100 employees when crossing $1.25B valuation
  • ~178 employees at Series C
  • Only 2 PMs at that scale

Compare to Atlassian at equivalent revenue: an order of magnitude larger.

product-development: The Linear Method

Linear published their methodology at linear.app/method — 8 principles:

  1. Build for creators (not managers/reporting dashboards)
  2. Purpose-built (not flexible catch-all)
  3. Create momentum, don’t sprint
  4. Meaningful direction
  5. Aim for clarity (no invented jargon)
  6. Say no to busy work
  7. Simple first, then powerful
  8. Decide and move on

Core primitives that replace Scrum vocabulary:

  • Cycles — auto-rolling 1-2 week time-boxes; unfinished work auto-carries (no sprint ceremony)
  • Triage — single inbox where all new bugs/feedback land before entering backlog
  • Projects — durable goals spanning cycles
  • Issues (not “user stories” — Linear explicitly rejects user-story ceremony)

No standups mandated; the product itself is the status meeting.

bootstrapping (Partial): Profitable and Independent

Linear isn’t bootstrapped — they raised $134M. But they operate with bootstrap discipline:

  • Profitable since 2021
  • Negative lifetime burn — more cash in bank than raised
  • No “fail fast” mentality — Saarinen: “Go slow to go fast”
  • Deliberate team size — refused to scale headcount with revenue
  • Small enough to keep craft — at $100M+ ARR with 100 employees, they maintained the Airbnb-design-team quality bar

This is the 37signals model with VC rocket fuel — capital deployed for runway, not growth-at-all-costs.

where-the-experts-disagree: The Contrarian Spectrum

Linear sits on the contrarian side of nearly every tension in the wiki:

TensionConventionalLinear
Marketing spend20-40% of revenue0.5%
LaunchOpen signups fast14-month invite-only
Free tierGenerous freemiumCapped, charge early
ExperimentationA/B test everythingZero A/B tests
CustomizationConfigurable for each customerOpinionated, rigid
Team sizeScale with revenueStay deliberately small
Move fastShip daily, break thingsCraft, zero-bug policy
PM ratio1 PM per 5-8 engineers2 PMs for all of Linear
HiringAlgorithm interviewsPaid work trials
Build in publicShare MRR publiclySelective; changelog-first

And every contrarian position was vindicated by the outcome.

Key Lessons

  1. Craft can replace marketing budget — when the product is screenshot-worthy, distribution is free. Design is the growth strategy.
  2. Opinionated beats flexible — customization is a feature for admins, not users. The tool that decides for you wins with people who have taste.
  3. Speed is a product, not a feature — 47ms vs 3.2s isn’t incremental; it’s categorical. Architectural decisions that enable sub-100ms interactions are a moat.
  4. Handpicked beta cohorts work — Saarinen personally selected ~10 users per week for 14 months. The slow path to 1,000 customers was also the most defensible one.
  5. Publish your methodology — The Linear Method turned adoption into ideology. Once you believe the philosophy, you can’t use Jira anymore.
  6. Profitable PLG is possible — charge from day one, cap the free tier, skip the generous-freemium trap. Linear proves PLG doesn’t require burning cash.
  7. 2 PMs for 15,000 customers — if engineers own product, most PM work is redundant. Linear’s hiring bar replaces the PM function with taste distributed across the team.
  8. VC money with bootstrap discipline — raising ≠ burning. Linear raised $134M but stayed profitable and kept more cash than they raised. Optionality without dependency.
  9. Twitter is a legitimate distribution channel — when the audience is engineers and designers, Twitter/X is where they live. The Sequoia partner who led Linear’s seed found them through her feed.
  10. The niche you avoid is as important as the one you pick — Linear refused to chase enterprise features for the first 2 years. The focus preserved the product.

See Also

Sources

Graph View

Backlinks