Karri Saarinen on Linear, the Linear Method, and Craft-Driven Product Development
Overview
Karri Saarinen co-founded Linear in 2019. The issue-tracking tool is profitable, carries net negative lifetime burn (more cash in bank than ever raised), runs on roughly 50 people, and is used by companies including Block, Vercel, and Ramp. The episode covers how Linear builds software — prioritising craft and fast internal iteration over upfront polish — how the company is structured without traditional product managers, and the published Linear Method that encodes these principles.
Key ideas
- Craft starts with hiring people who care, not with process. Karri argues quality cannot be procedurally imposed on people who do not intrinsically value it. The prerequisite is hiring engineers and designers who feel personal ownership over what ships. Linear tests for Product Taste in engineering interviews by asking candidates to evaluate real product decisions: ‘Why was this decision made? Do you agree with it?’
- Push janky internal versions fast; fix fast. Counterintuitively, the craft-driven path runs through fast internal shipping rather than perfecting before release. Visible internally only → feedback from 1–5 customers → polished general release. The speed of the feedback loop is the quality mechanism, not the thoroughness of pre-release review.
- The Linear Method gives teams good defaults rather than infinite flexibility. Opinionated software removes the meta-work of configuration. Cycles automate the schedule so teams focus on agreed priorities rather than backlog negotiation. No OKRs or per-feature metrics — quality is measured by feel and customer empathy.
- One head of product; engineers and designers own projects end-to-end. Linear has one PM (Nan Yu). Project leads — who may be engineers or designers — carry the PM responsibilities for their project. This demands high-calibre generalism and forces the company to test for product sensibility at the engineering and design hiring stage, not just technical skill.
- Product-market fit is a spectrum, not a moment. Linear found PMF with early-stage startups first, then expanded. Rather than targeting new segments proactively, they followed pull: AI companies and crypto companies gravitated to Linear, so the company doubled down on those communities.
Craft and quality
Karri’s account of quality inverts the common intuition. The path to a polished product runs through speed of iteration, not thoroughness of pre-release review.
The sequence Linear uses:
- Ship a rough version to internal users only. Visible inside the company; not customer-facing.
- Collect feedback from 1–5 real customers. Small sample, direct conversation.
- Polish and ship the general release.
The roughness of step one is intentional. Getting something in front of real users quickly surfaces the actual problems rather than the problems the team imagines. Over-investing in polish before external contact is, in Karri’s view, a waste of the resources that should go into the iteration cycle.
Product Taste is the underlying capacity: the ability to know what good feels like, to identify what is wrong with an early version, and to know when the iteration cycle has produced something worth releasing. Karri tests for this in engineering hiring — asking candidates why specific product decisions were made, whether they agree, and what they would do differently — because taste cannot be taught as easily as technical skill.
The Linear Method
The Linear Method is Linear’s published set of principles for modern software development. Its core commitments:
Opinionated software. Rather than offering maximum flexibility, Linear makes choices for users. The reasoning: time spent configuring is not time spent working. Good defaults collapse the decision space so teams spend their attention on the actual work. This is a deliberate product philosophy, not a limitation.
Cycles, not sprints. Cycles are like sprints but with automated scheduling. The cadence is not a management ceremony; it is a coordination device. The key distinction from traditional sprint rituals: cycles help teams focus on agreed priorities rather than letting the backlog grow without bound and attention shift to whatever is loudest.
No OKRs, no per-feature metrics. Linear does not track output by feature. Quality is assessed by feel — does the product work well? — and by direct customer empathy. Karri’s view: metric-driven development optimises for what is measurable rather than what matters. The company trusts its own judgment and customer signal over dashboards.
Project leads, not PMs. Engineers and designers take project lead roles, owning the problem definition, the trade-offs, and the communication — responsibilities traditionally assigned to product managers. One head of product (Nan Yu) sets direction. Everyone else executes with ownership.
Hiring and team structure
Linear’s structure forces an unusually high hiring bar on every role.
No PMs except one. Because project leads take PM responsibilities, every engineer and designer hired must be capable of product reasoning, not just execution. Karri tests this directly in interviews: present a real product decision, ask why it was made, ask whether the candidate agrees. The answer reveals whether the person has an opinion about the product — the minimum bar for project ownership.
Paid work trial. The final hiring step is a paid contractor assignment. The problem statement is deliberately vague; the candidate gets access to the codebase, Slack, and Notion. What Karri watches for:
- Does the candidate make progress without being told exactly what to do?
- How do they approach a problem where the right answer is not given?
- Do they take ownership, or do they wait for direction?
The trial reveals operating style under realistic conditions — more signal than any interview question.
YC’s core advice as a hiring and prioritisation filter. Karri cites YC’s canonical frame — talk to customers, build the product, exercise; everything else is a side quest — as a practical filter for all decisions. Linear applies it to hiring too: does this role serve the main quest (better product for users), or is it a side quest?
Growth
Private beta cohort model. Linear ran a private beta for a year before public launch. Rather than opening access broadly, they onboarded roughly 10 users per week, collected feedback, fixed problems, and invited the next cohort. Several hundred companies were on the product before public launch. Conversion to paid after launch was close to total.
Pricing experiment. Rather than setting a price based on competitive benchmarking, Linear offered a sliding scale from $0 to $20 per seat and let early users choose. The purpose was to test willingness to pay — to understand what users thought the product was worth — rather than to maximise early revenue.
Following pull into new segments. Linear found PMF with early-stage startups first. Rather than engineering expansion into enterprise or other verticals, the team watched which communities gravitated to the product unprompted. AI companies and crypto companies showed the strongest pull; Linear doubled down there. The lesson Karri draws: PMF is not binary and not uniform — it manifests differently in different segments, and expansion follows signal rather than planning.