The email from Mark Zuckerberg arrived at a small Amsterdam studio in 2011. It invited Koen Bok and Jorn van Dijk, the founders of Sofa, to come to Palo Alto to discuss a “future partnership.”
When they arrived, it was clear the conversation was about an acquisition.
Sofa was a 15-person design studio that had won two Apple Design Awards in consecutive years, one for Checkout and one for Versions. They had made beautifully crafted Mac applications, worked with clients including TomTom, Mozilla, and Nike, and built a reputation for the kind of software that felt intentional down to the last pixel. Facebook at the time was scaling engineering headcount rapidly but had few designers relative to engineers. Zuckerberg wanted design capability. Sofa was exactly the kind of team he was looking for.
Koen and Jorn took the deal. They moved to Menlo Park. Koen worked on Messages, Ads, and Video. Jorn worked on mobile apps and Photos. They reported to Zuckerberg. They spent over two years learning how a company of Facebook’s scale builds and deploys products, absorbing lessons about growth, product development, and organizational design that a 15-person Amsterdam studio could not have taught them.
By late 2013 they were done. They had learned what they came to learn. They left Facebook with a problem they wanted to spend the next ten to twenty years on: helping companies become digital-first and product-led. They spent several months experimenting with different directions before landing on the one that became Framer.
The product they eventually built, abandoned, rebuilt, and pivoted was in each version trying to answer the same underlying question: why is there still a painful gap between what a designer imagines and what a developer ships, and what would it look like if that gap didn’t exist?
Sofa, Facebook, and the Design Education That Followed
Jorn van Dijk studied graphic design at design school, then attended art academy in the Netherlands, absorbing the Dutch tradition of typography and systems thinking that had produced influential designers for decades. He was part of a generation where government institutions had invested heavily in forward-thinking design education, and the influence showed in how he thought about software. Not as engineering with a visual layer added afterward, but as a unified craft where the interaction, the visual form, and the underlying system were inseparable.
His path into product software came from finding people who were obsessed with the same intersection of interface and craft. He joined Sofa as its first employee, working alongside Koen Bok, who had started the studio. Together they built products that demonstrated what it looked like when software was treated as a design problem first. Checkout won its Apple Design Award for the quality of that approach applied to a transaction product. Versions won its award the following year.
The Facebook acquisition was not primarily a financial event for them, though it obviously was that too. Koen and Jorn framed the move explicitly as an educational opportunity: they wanted to understand how Silicon Valley companies built products at massive scale, and joining Facebook was the fastest way to get that education in depth. The culture shock of going from 15 people to tens of thousands was real, but it was intentional.
When they left in 2013, they brought back a specific perspective. At Facebook’s scale, the gap between design and implementation was not an incidental friction but a structural bottleneck. The cycle of designing something in Photoshop or Sketch, handing it off to engineers, watching it come back different, revising, re-handing-off, iterating through that cycle repeatedly, consumed enormous amounts of time and created real degradation of design intent between concept and shipped product. The problem was systemic.
They spent nearly a year after leaving Facebook experimenting with different approaches to solving different problems. Eventually they arrived at the prototyping tool that became Framer’s first product.
Framer Classic and the Ceiling That Prototyping Hit
The original version of Framer was built on a JavaScript framework Koen had been developing, initially released as open source, that let designers animate and prototype interfaces using CoffeeScript code. For designers who were willing to write code, it was unusually powerful: you could express precise interaction behaviors, complex state transitions, and sophisticated animations that static design tools couldn’t capture and that basic prototyping tools couldn’t replicate.
In 2015, they released Framer Studio, a desktop application for macOS that added a visual canvas alongside the code editor, reducing the barrier for non-coders while keeping the scripting power for complex interactions. The product supported importing designs from Sketch, layering vectors and images, and defining prototypes with taps, swipes, and scrolls, outputting shareable links that showed the prototype running in a browser.
Framer Classic spread quickly. Designers at Dropbox, Twitter, Microsoft, and Amazon adopted it. The reputation was strong in the exact community that Koen and Jorn had targeted: senior product designers at technology companies who were frustrated with the limitations of static mockups and basic hotspot prototypers. These were people who either knew how to code or were willing to learn, who cared intensely about the fidelity of the prototypes they used to communicate with engineers and stakeholders.
The product worked. Revenue grew. And then it plateaued.
Jorn described the plateau with unusual directness in interviews: they had built a prototyping tool and believed that high-fidelity interactive prototyping would eventually replace static mockups across the industry. In reality, most design teams kept treating prototyping as optional. The workflow that Framer Classic enabled was genuinely superior but required effort that most teams weren’t willing to invest. They tried education and evangelism for roughly a year, hoping adoption would accelerate. It didn’t.
The prototyping market had a ceiling. By the time Framer’s revenue peaked around $4 to 5 million ARR in 2019, it had become clear that ceiling was not significantly higher than where they were. They had done excellent work reaching it. The work hadn’t been wasted. But the product category they had chosen was not going to become a large business on its own.
Framer X and the Second Bet That Also Didn’t Stick
In 2018, after their Series B of $24 million led by Atomico, Framer launched Framer X. The bet this time was different: instead of a prototyping tool that required code, Framer X was a fully integrated design and prototyping environment powered by React, capable of importing production components and assets from live codebases and using them directly in design and prototype work.
The vision was compelling. If you could bring real production React components into your design tool, the gap between the prototype and the shipped product would collapse. Designers could build prototypes from the same components engineers were shipping. The handoff problem, the thing Koen and Jorn had been trying to solve since Sofa, would be solved at the component level.
The product was technically impressive. But it introduced new frictions. Most designers did not work in React and did not have a mental model for component-based design. The teams that were most excited about Framer X were technically sophisticated product designers at engineering-led companies, a narrower audience than the broader design market. The COVID-19 downturn in 2020 hit the prosumer design tool market hard. Revenue declined from its peak.
After Framer X, Framer went into what Jorn described as a “make or break” moment. Growth had stalled after the Series B. Team members left. Koen and Jorn shifted roles: Koen leaned into product, Jorn took on commercial responsibility. They went back to the core question with fresh willingness to tear everything down and rebuild if necessary.
The question they asked themselves was direct: could their design tool build actual production websites?
The Pivot: One Question That Changed the Company
In 2021, Koen and Jorn looked at the Figma-to-Webflow workflow that professional designers and marketing teams were navigating as their standard process for building websites.
The workflow went like this. A designer created a site design in Figma, the dominant design tool by that point. When the design was approved, they had to rebuild it from scratch in Webflow or have a developer code it, because Figma files did not translate to live websites. Every iteration on the design required another round of rebuilding. The handoff between design and implementation happened manually, requiring either a developer partner or a significant investment of Webflow-specific technical learning.
This was not a niche frustration. It was the standard workflow for building marketing websites, landing pages, portfolio sites, and brand pages across the industry. Millions of designers and marketing teams were navigating this broken loop daily.
What if Framer collapsed it? What if the design canvas and the published website were the same thing?
The team spent roughly nine to ten months executing the pivot after acknowledging it was necessary. They rebuilt Framer around the goal of making design and publishing a single unified action. You design the site in Framer. You hit Publish. The site is live. No rebuild step. No handoff. No Webflow.
They launched the website builder publicly in May 2022. By December of the same year, they had reached $1 million ARR.
From $2.5 million ARR at the bottom of the prototyping decline to $1 million in new revenue in seven months from the new product. The market feedback was immediate. By 2023, Framer was at $10 million ARR. By 2024, $25 million. By mid-2025, $50 million. The growth rate had approximately doubled each year since the website builder launched.
Why the Design-First Positioning Works
The question that Framer gets asked in the same way Webflow gets asked about Wix is: how are you different from Squarespace?
Squarespace and Wix are template-first tools built for people who want a website quickly without caring much about the underlying design. They optimize for accessibility to non-designers. The aesthetic output is competent but constrained by the template structure. You cannot make a Squarespace site look like it wasn’t built on Squarespace.
Framer is built for designers who have opinions. The interface is familiar to anyone who has used Figma: it is a real design canvas with real layout controls, real typography tools, real animation and interaction capabilities. You are not selecting from templates and filling in content. You are designing the site from its visual structure up, with direct access to the actual CSS layout properties, the actual typography scales, the actual motion design parameters.
The output is different in a way that is visually legible. Framer sites tend to look distinctly designed rather than distinctly templated. The animations are smooth because the platform is React-based and generates performant code rather than the bloated output that some visual builders produce. SEO performance is built in because the generated HTML is semantic and clean. Global CDN hosting is included because Framer controls the deployment infrastructure.
The co-founders described the product with reference to a specific pain point: designers often create web designs in Figma, only to have to rebuild them from scratch in web development platforms like Webflow. With Framer, they eliminated that. Design your site, make it responsive, add animations, and publish, all in one place. No more endless back-and-forth between tools.
The designers who adopted Framer after the May 2022 launch were not looking for a simpler tool. They were looking for a better one that respected their design sensibility and removed the implementation friction that was the main cost of their current workflow.
The Freelancer-to-Enterprise Transition
Framer’s initial wave of adoption after the 2022 launch followed a similar pattern to Webflow’s early growth: individual designers and freelancers who used it for client work, liked it, and introduced it to the organizations they worked with.
The Figma-like interface was the specific mechanism that made this adoption fast. Designers who were fluent in Figma could open Framer and immediately orient themselves. The concepts, frames, auto-layout, components, variants, were familiar. The difference was that hitting Publish produced a real live website rather than a design file that still needed to be rebuilt somewhere else.
For freelancers building marketing sites and portfolios for clients, the efficiency gain was immediate and substantial. A project that previously required designing in Figma and then rebuilding in Webflow, effectively two full rounds of work, collapsed into one. The client received a site faster. The designer charged more per hour because they were delivering the complete product rather than half of it.
As with Webflow, each freelance customer was also a distribution channel. The sites they shipped on Framer were built on Framer infrastructure. Clients who needed to update those sites became Framer users through the relationship with their designer.
The enterprise shift happened more recently and more deliberately. Framer introduced business plans with enterprise-grade features: advanced analytics, A/B testing tools, enterprise security, SSO, audit logging, advanced CMS capabilities for dynamic content at scale. These were not features individual freelancers needed. They were features that the marketing teams at real companies needed to run their actual marketing website infrastructure on Framer rather than on developer-maintained custom codebases.
The enterprise segment grew rapidly once those features existed. Since launching business plans, the majority of new Framer customers have been businesses rather than individuals. Miro, Perplexity, Scale AI, and Bilt Rewards joined. Forty percent of Y Combinator’s most recent batch selected Framer for their marketing websites.
The Y Combinator penetration is meaningful for the same reason it is meaningful for PostHog and Webflow: YC founders are technical, skeptical of vendor lock-in, and deeply networked. When 40% of a YC batch independently chooses the same website platform, it travels through the network. Every founder who recommends Framer to a portfolio company from another batch, every investor who sees consistent Framer adoption and starts recommending it to companies they fund, every YC alum who takes their Framer workflow to the next company they join or found is a distribution event that Framer did not pay for.
The Technical Architecture and Why It Matters
Framer is built on React. This is a specific technical choice with downstream consequences for the quality of the product and the competitive moat it creates.
React is the dominant front-end framework for professional web development. The websites that professional engineering teams build at real companies are overwhelmingly built with React or React-adjacent frameworks. When Framer generates the code for a website you design on its canvas, it generates React code, the same code a skilled front-end engineer would write.
This has two important implications. First, the output of Framer sites performs well. The generated code is clean, semantic, and production-grade. SEO performance is strong because the HTML structure is correct. Page load times are fast because the React rendering is efficient. Animations are smooth because they’re implemented with proper React motion libraries rather than CSS hacks.
Second, it is possible for developers to work on Framer sites without feeling like they’re fighting an abstraction layer. A developer who needs to add custom behavior to a Framer site can drop into the code, make changes, and have those changes respected by the design canvas. The design tool and the codebase are not fundamentally different representations of the same thing. They are the same thing expressed two ways.
This is what Framer means when it describes itself as a design canvas that publishes directly. It is not a website editor that generates export files that then need to be deployed. It is a design environment whose output is the actual deployed website.
The AI layer that Framer added to the product fits naturally into this architecture. When you prompt Framer AI to generate a page or section, it draws from a library of approximately 100 pre-designed sections built on the same design system and generates a starting point that a designer can refine from. The result is described internally as a reliable “7/10 first pass”: not finished, but correct enough structurally and visually that the designer is polishing rather than rebuilding. The quality bar of the AI-generated output is higher than competitors whose AI generates generic layouts because Framer’s design system has a specific visual identity that the training reflects.
The Revenue Trajectory and What It Says About the Pivot
The revenue numbers tell the story of the pivot better than any narrative description.
Framer peaked at approximately $4 to 5 million ARR with the prototyping tool, declined to roughly $2.5 million ARR during the COVID period, held at low single-digit millions through the Framer X years, launched the website builder in May 2022, reached $1 million in new ARR by December 2022, grew to $10 million ARR in 2023, $25 million in 2024, and $50 million by mid-2025.
The doubling each year from the 2022 website builder launch is the signal. That kind of growth rate, sustained over three consecutive years, is evidence of genuine product-market fit rather than a one-time spike. Framer had found a market that wanted what it was building, at a price those customers were willing to pay, with growth rates that could compound into something large.
The valuation trajectory followed the revenue: $27 million Series C in 2023 at an undisclosed valuation, $100 million Series D in August 2025 at a $2 billion valuation. Total funding raised across all rounds reached $161 million. The company has been break-even for the past year. The $100 million Series D is growth capital rather than survival capital.
Koen Bok told TechCrunch at the Series D announcement that Framer aims to cross $100 million ARR in 2026. The goal is doubling from $50 million in one year, which would require roughly the same growth rate the company sustained in 2024 and early 2025. The enterprise B2B shift, with its higher average contract values and lower churn relative to individual subscriptions, is the mechanism management is betting on to hit that number.
The Broader Story This Sits Inside
Koen and Jorn’s original goal when they left Facebook in 2013 was to work on something for ten to twenty years. Most founders say things like that in pitch decks. Few mean it operationally. The evidence that they meant it is the eleven years of building and rebuilding that actually followed.
The prototyping tool was not a distraction from the real thing. It was the first serious attempt at solving the design-to-implementation gap, tested in the real market, found to have a ceiling, and iterated past. Framer X was the second attempt, better in some ways, still not the right answer. The website builder was the third version of essentially the same underlying idea: why can’t a designer build the actual thing?
The answer they eventually found was that they could, if the design canvas and the published site were unified rather than separate. That insight required building three versions of the product and watching two of them plateau to understand clearly enough to get the third version right.
The Apple Design Awards at Sofa established a standard of craft. The Facebook experience established an understanding of scale. The prototyping years established a deep technical knowledge of what browsers could do and what React-based rendering could produce. Each phase contributed something to the version of Framer that works.
Jorn van Dijk describes Framer’s position now with a clarity that reflects how long he has been thinking about this specific problem: for more than a decade, tools like Wix and Squarespace made it easier to spin up personal sites. But for professional, high-traffic, brand-defining websites, teams were still stuck in developer-heavy workflows. That was the gap Framer was closing.
500,000 monthly active users. $50 million ARR. $2 billion valuation. Break-even operations. 40% of the latest YC batch. Miro, Perplexity, Scale AI.
The gap they identified in a small Amsterdam studio in 2013, after two years at Facebook and a decade before that building software they were proud of, turns out to have been the right gap to close. The market has been waiting for it since the web was invented. They got there on their third try.
Leave a Reply