Figma 2026 Q3 Earnings Call
Review the key takeaways and the transcript of this earnings call.
- Q3 revenue was $391 million, up 43% year over year and $16 million above the high end of prior guidance.
- Net dollar retention rate was 131%, marking the 5th consecutive quarter above 130%; gross retention rate remained in the mid to high 90s.
- Non-GAAP gross profit was $321 million, up 36% year over year, with an 82% gross margin; non-GAAP operating income was $47 million, a 12% operating margin.
- Free cash flow was $42 million, a margin of 11%, and Figma ended Q3 with $1.7 billion in cash, cash equivalents, and marketable securities.
- Figma added more than 600 paid customers with more than $10,000 in ARR and more than 80 paid customers with more than $100,000 in ARR in Q3; paid customers with more than $10,000 in ARR represented nearly 70% of ARR at quarter-end, compared with 64% at the end of Q1 2025.
- The Figma MCP server had approximately 1 million weekly active users; total MCP usage grew 3x quarter over quarter and writeback usage grew 4x in Q3.
- Figma Agent was generally available on October 6th. As of October 31st, 65% of paid customers with more than $10,000 in ARR were using the agent weekly, and over 80% were consuming AI credits.
- From the start of Q3 through October 31st, weekly AI credit consumption increased 2.5x, and Figma more than doubled the share of work powered by its own post-trained design model.
STOCKNOW INSIGHTS
Continue with outlook and guidance.
Log in to unlock executive comments and Q&A highlights.
Log in for the full summaryStockNow uses AI to translate and summarize earnings calls. Accuracy and completeness are not guaranteed.
Transcript
Preview the first fifteen paragraphs, organized by speaker.
Hello, everyone. Thank you for joining us, and welcome to the Figma third quarter 2026 earnings call. After today's prepared remarks, we will host a question-and-answer session. If you would like to ask a question, please press star one to raise your hand. To withdraw your question, press star one again. I will now hand the conference over to Kate DeLeo, Vice President of Investor Relations.
Kate, please go ahead. Good afternoon, and thank you for joining us on today's conference call to discuss Figma's results for the third quarter of 2026.
On the call, we have Dylan Field, Figma's Co-founder and Chief Executive Officer, and Praveer Melwani, our Chief Financial Officer. During the course of today's call, we may make forward-looking statements, including but not limited to statements regarding our guidance and future financial performance, market demand, product development, growth prospects, business strategies and plans, partnerships, ability to attract and retain customers, and ability to compete effectively. These forward-looking statements are based on management's current views and assumptions and should not be relied upon as of any subsequent date, and we disclaim any obligation to update any forward-looking statements. Actual results may vary materially from today's statements.
Information concerning our risks, uncertainties, and other factors that could cause results to differ from these forward-looking statements are included in our filings with the SEC, including our quarterly report on Form 10-Q for the quarter ended September 30, 2026. Our discussion today will include certain non-GAAP financial measures. These non-GAAP financial measures should be considered in addition to, not as a substitute or in isolation from, GAAP measures. Our non-GAAP measures exclude the effects of our GAAP results of stock-based compensation and certain other items. Reconciliations of non-GAAP financial measures to comparable GAAP measures can be found in our press release accompanying this call, which is posted to the investor relations page on our website. I would now like to turn the conference call over to Dylan.
Hi, everyone, and thank you for tuning in to today's call. For decades, writing code was the bottleneck of the product development process. As coding models improve, that is no longer true. Value is moving up the stack to design, and demand for great design is growing. According to our research, the majority of product builders, developers included, believe that design skills are becoming more important than coding skills. We see this in how teams use AI in Figma, with our own agents and others that are connected to our canvas. To design something well, AI needs the context that lives on Figma's platform, your thinking, your design system, the decisions that your team has made. Design is the new code, and Figma is the center of gravity for design. This expands Figma's market opportunity, and we're already seeing it in our results.
I'll walk through a few highlights from the quarter. In Q3, we delivered $391 million in revenue, up 42% year-over-year, and $16 million above the high end of our prior guidance. Our net dollar retention rate was 131%. Non-GAAP gross profit dollars grew 36% year-over-year. Our non-GAAP operating margin was 12%, and our free cash flow margin was 11%. We finished Q3 with $1.7 billion in cash equivalents, and marketable securities. We delivered these results as we focused on building the full-stack canvas for visual-first development. It's made for collaboration, increasingly powered by our first-party model, and native to the ways that teams design and build with AI. This is Figma's opportunity and what we believe will fuel our next phase of growth ahead. We're tackling this opportunity by extending the Figma platform on three main fronts. First, code and context. Second, the Figma Agent and the Figma MCP server.
Third, our first-party model. Let's start with code and context. Earlier this year, we introduced Code Layers, which launched into open beta last month. As code becomes a commodity, more designers are using it to visualize their thinking and bring their ideas to life. That work is starting with coding agents in terminals and IDEs. These tools are powerful, but working in a linear, code-first way can lead to tunnel vision and siloed workflows. When you're working alone, it's easy to get attached to your first idea and build the wrong thing, which ends up slowing everyone down. We see this in our research. More than half of product builders, 52%, said fixing AI outputs is becoming one of the most time-consuming parts of their day-to-day work.
As one customer told us, it's, quote, "Fast, fast, fast, and then slow." We also know that for many of our customers, and basically anyone who's not a developer, the terminal, the IDE, they just don't feel like home. They prefer the canvas, where they can explore and compare and riff on ideas with their team. Agents make the canvas even more valuable. The faster AI can help you generate prototypes, the more teams need a shared place to see them side by side to decide which ones are worth developing, which deserve to ship. This is where Code Layers shines. At Merkle, teams are reporting 30% faster iteration and feature development with Code Layers.
As Merkle's VP and Global Lead of Experience Design, Dave Calleja, put it, "The power of Code Layers is removing the perceived division between code-based and canvas-based tools with the full power of both." The design, engineering, and product teams at Cox Automotive are using Code Layers to explore variations on a new dealer-facing CRM. A Senior PM for AI went through 19 versions of that CRM with about an hour of hands-on keyboard time. In their words, working with Code Layers was, quote, "At least 10 times more efficient or about 90% time saved." Code Layers rolled out to all users this week. We see it as core infrastructure that we'll continue to invest in and improve. It's already changed how we build at Figma, and we're excited about Code Layers' potential to transform how teams build with code in a visual-first way.
In addition to code, we are also introducing ways to bring more context into the Figma canvas with Specs. To understand the power of Specs, it helps to understand Markdown, a context format that has become widely used to feed context to agents. Specs is a medium for Markdown that lives directly on the Figma canvas. It gives teams a new way to codify context and move it into, out of, and across Figma. It works with all content types, visual, written, design, and also code. It is also multiplayer, meaning you can work on Specs with both human and agent collaborators. Specs houses teams' shared decisions and can reference work already in the canvas, giving teams a better way to ultimately ship their design intent. That brings me to the second area of focus, agents, starting with the Figma MCP server.
Figma's MCP server has become an important way of how teams build with AI. Recently, we have added several new capabilities, including the ability to generate Code Layers, images, or push a Markdown file into Figma Specs. Last month, we added 20 new MCP partners so that workflows can easily connect to our platform. These new partners include Notion, Xcode, and Meta's AI agent, Muse, which was also designed in Figma. During Q3, the Figma MCP server was our fastest-growing product by growth in weekly active users. Today, it has approximately 1 million weekly active users. In Q3, total MCP usage grew 3X quarter-over-quarter, and writeback usage grew 4X. This reflects Figma's central place in the product development workflow and the power of working on the canvas in a visual-first way. MCP is also bringing new teams to Figma.
In Q3, about half of new Pro teams used our MCP server, and customers using Figma MCP grew full seats faster than those who did not. It is also translating into meaningful AI usage on our platform. In fact, MCP is now driving 5% of credit consumption on Figma. Turning to the Figma Agent. The agent is foundational to the full-stack canvas we are building. Customers are using the agent to generate designs grounded in their design systems, to explore ideas and get them in front of users faster, and to save time by automating repetitive tasks. For example, at Uber, the agent helps document design system components across seven different platforms. Documentation that once took multiple people months to complete can now be published by one designer in an afternoon. At Granola, the Figma Agent uses the Granola connector to turn meeting feedback into annotations directly on the Figma canvas.
That way, designers can stay focused on the conversations they are having and then act on what they heard. As one of Granola's designers put it, "The quicker you can get a design in front of a user, the better. Everything I do with the Figma Agent is in service of that." As of October 31st, on a weekly basis, 65% of paid customers with more than $10,000 in ARR were using the Figma Agent, and over 80% were consuming AI credits. Across Figma, excluding beta usage, weekly AI credit consumption has increased 2.5X through the transition of the Agent GA launch in October 31st. The Figma Agent and MCP server are also bringing new personas to Figma that want to work in a design and visual-first way. We believe this expands our total addressable market beyond traditional designers and product teams.
For example, at Harvey, the marketing team is using agents connected through the Figma MCP server, as well as tools like Figma Motion and Figma Weave, to create launch films and personalized product videos. This is part of what they're calling a, quote, "frontier marketing org." They built reusable libraries for their brand guidelines, product screens, and motion animations, all in Figma. This gives agents the context to build consistently and at the quality bar that their team expects. The team visually reviews storyboards on the canvas while agents revise the work based on their feedback. In one batch, Harvey produced 36 films in 69 minutes. It's a workflow where human direction, shared design context, and agents come together to, as they put it, quote, "turn ambitious ideas into repeatable work at scale." The third area of focus is the first-party model that's powering more of the agent's work.
Growing usage of the Figma Agent raises the bar for how our users expect it to perform. Every edit needs to be useful, precise, and fast enough to keep people in their flow. This is why we're investing in our own first-party models. We've seen that post-training with expert designers and researchers calibrating our evals around design can compete meaningfully with frontier models on specific tasks. From the start of Q3 through October 31st, we more than doubled the share of work powered by our own post-trained design model. Along with third-party models, it's delivering a better experience for our users. In head-to-head testing with expert designers, the agent beat or matched the previous production agent 62% of the time. User ratings are 33% higher, and users undo the agent's work 60% less often. We're excited about the potential of this work for two main reasons.
FULL TRANSCRIPT
Continue the full translated transcript in StockNow.
Log in to unlock every statement, the English original, and speaker-by-speaker history.
Log in for the full transcriptCall participants
16 people spoke on this call — only 2 are shown here.
PARTICIPANT LIST
View participant details in StockNow.
Log in to see executives and analysts, their roles, and complete speaking history.
Log in to view all participantsKeep exploring
