May 16, 2025 · 1h 1m · 20product
20Product: Figma CPO on How Figma Builds Products: What Works, What Does Not | How Figma Does Testing and Product Reviews | The Future of Design, Engineering and Product with Yuhki Yamashita
0:00 / 0:00
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of 20 Product, host Harry Stebbings interviews Yuhki Yamashita, Chief Product Officer at Figma, to explore Figma's product development processes, design philosophy, and organizational alignment strategies. Yuhki shares insights on harnessing user hacks, leveraging AI to redefine cross-functional roles, and maintaining human craftsmanship in modern software design.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Harry holds 32.6% of the talking time here. How this is scored →
Harry as informed peer 4.9 Guest teaching 3.8 Guest disagreement 1.5 Harry pushing back 3.4
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 23:20 Parrying engineer headcount reduction prompts
When Harry repeatedly presses Yuki to admit AI will reduce engineering pod headcounts, Yuki playfully resists the premise, remarking that Harry is in the business of speculative guessing before asserting Figma won't cut engineers.
Hardest push from Harry ▶ 32:05 Challenging the necessity of Product ManagersHarry directly challenges the core premise of PMs, arguing that the founder owns the 'why' and asking if Figma could function without product managers altogether.
Biggest teaching moment ▶ 38:04 Dylan Field's Design Twitter distribution graphYuki educates Harry on how Figma achieved early viral distribution by mapping Design Twitter as a network graph and directly targeting key node influencers rather than relying solely on organic product discovery.
Harry holds his own ▶ 36:50 Quoting Spotify's product philosophy on craft vs speedHarry demonstrates high product expertise by quoting Spotify's Gustav Soderstrom ('the details are the product') to frame a sharp question on balancing shipping velocity with craftsmanship.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Upbringing, Questioning Assumptions, and Simplicity in Design | 4 | 3 | 1 | 3 | Harry challenges the conventional wisdom that 'simple is always better', asking whether simplicity can become lazy design versus feature creep. Yuki explains that simplicity is a perception and describes how Figma offers dual modalities like Design Mode to serve power users without overwhelming novices. | |
| Inside Figma's Internal Pitch Process and Storytelling | 3 | 3 | 1 | 2 | Harry asks about Figma's internal pitch and roadmap process. Yuki shares how organic user hacks (such as creating millions of slide decks in Figma Design) inspire new products like FigJam and Figma Slides. | |
| The Evolution of PRDs and Metric Alignment with KPI Trees | 5 | 4 | 2 | 4 | Harry provocatively asks if PRDs are dead and challenges Yuki's 'screenshot test' concept by asking about complex products that cannot be captured in a simple GIF. Yuki clarifies how PRDs have evolved into centralized North Stars and introduces KPI trees. | |
| Multiple Entry Points and Real-Time Code Generation | 5 | 3 | 1 | 3 | Harry probes the future of product entry points, comparing prompt-to-code builders like Bolt and Lovable against traditional design-first tools. Yuki articulates Figma's philosophy of offering multi-modal entry points for different builder personas. | |
| Redefining Design and the Convergence of Disciplines | 4 | 3 | 1 | 2 | Harry notes the collapsing of functional roles into 'superhumans'. Yuki explains how Figma Make shifts the underlying source of truth to code while expanding the broader definition of design. | |
| Future Team Pod Structures and AI-Assisted Engineering | 5 | 4 | 3 | 4 | Harry tries to force Yuki into predicting reduced engineering headcount ratios in future pods due to AI leverage. Yuki gently parries, pointing out that Harry is in the business of making speculative guesses while maintaining that Figma will leverage AI to do more rather than downsize. | |
| Evaluating Product Constraints and Canvas Dimensions | 5 | 4 | 1 | 2 | Harry demonstrates domain reflection by sharing a personal misstep about maintaining an audio-only constraint for his podcast. Yuki explains how the 2D infinite canvas is liberating for designers but confusing for non-designers who require 1D linear workflows. | |
| Learning from Product Failures and AI Model Transparency | 5 | 4 | 1 | 5 | Harry catches a potential contradiction, asking if explaining underlying AI models contradicts the screensaver simplicity discussed earlier. Yuki reflects on the public backlash to Figma's 'Make Design' release and the necessity of transparent model attribution. | |
| The Essential Role of PMs and Strategic Storytelling | 6 | 4 | 3 | 6 | Harry strongly challenges the PM discipline, asking if the 'why' belongs to the founder and whether Figma could function without PMs entirely. Yuki defends the necessity of PMs as objective mediators balancing engineering feasibility and design vision. | |
| Craftsmanship, Playfulness, and Balancing Speed with Quality | 6 | 3 | 1 | 4 | Harry cites Gustav Soderstrom ('the details are the product') and Johnny Ive to ask how Figma balances shipping urgency with extreme craft detail. Yuki emphasizes shipping fast to validate critical assumptions before polishing speculative details. | |
| Measuring User Reliance, Distribution Strategies, and Root Needs | 4 | 5 | 1 | 2 | Yuki details Figma's core reliance metric (L7/ND7) and educates Harry on Dylan Field's early distribution strategy of mapping Design Twitter as a influence graph. Yuki also explains the '5 Whys' method for uncover underlying user needs. | |
| Scaling Figma's Product Suite and Managing Tool Betas | 6 | 4 | 3 | 6 | Harry pushes hard on Figma launching four products simultaneously rather than drip-feeding them, and compares delayed product shutdowns to late employee terminations. Yuki politely declines to name internal product flops publicly. | |
| Bridging Design to Code and Beating Team Product Funks | 5 | 4 | 1 | 4 | Harry asks whether the disconnect between design mockups and production code is a feature or a bug. Yuki explains why it is primarily a bug caused by designers designing too far ahead of production reality. | |
| Resource Allocation, AI Differentiation, and Lessons from Uber | 5 | 5 | 1 | 2 | Yuki addresses resource limits and AI differentiation through craft. He shares a historical case study from Uber where CEO Travis Kalanick pushed upfront pricing despite internal fears of cannibalizing profitable Uber Black revenue. | |
| Quickfire Round: Figma Breakups, Leadership Weaknesses, and Waymo | 5 | 4 | 1 | 2 | In the quickfire round, Yuki shares how Figma rallied post-Adobe merger collapse around their core mission, critiques his own stoic leadership style, and praises Waymo's subtle UX details. |
Statements from this episode (35)
Yamashita: Over-Prioritizing Simplicity Stifles Necessary Feature Growth
“So there is truth to that, but simplicity is a perception. So it's not necessarily a reality. You know, if you take simple is always better to an extreme, you might not add new features. You might not add new capabilities. And I don't think that's always the r…”
Yamashita: Observing User Hacks Reveals Real Signal for Product Roadmaps
“It's important to observe how people hack your product, you know, use that as inspiration, because that's real signal, right? They're going out of their way to demonstrate to you that they have a need, even though you're not delivering on it.”
Yamashita: Users created 3.5M presentation decks in Figma before Slides launched
“Before our Slides product, three and a half million decks were being made in Figma design, even though it wasn't intended for it.”
Yamashita: Tangible prototypes convince stakeholders far better than theoretical pitches
“Building something tangible that people can use convinces people more that this idea is tenable.”
Yamashita: Prototyping with new capabilities must be a core product team skill
“Well, I think it's really important right now for every product team to think about how to prototype and play with new capabilities because it's, everything is so fast moving, so programs like Maker Week help, but I think it's a core capability that we all nee…”
Yamashita: Uber structured all team KPI trees under trip count
“At Uber, at the very top was trips. Everything cascaded from there.”
Yamashita: Products requiring tutorials to explain value signal design failure
“All too often people end up kind of having to walk people through, okay, this tutorial of like, let me just explain to you why this product is better. And to me, that's not just a story. It's also kind of a constraint in the design. Like you're maybe not pushi…”
Stebbings: Startups frequently fail due to choosing the wrong entry point
“Often you lose because you get the wrong entry point. You have the right idea, the right solution, but the wrong entry point.”
Yamashita: Future product building tools must support multiple entry points
“The future has to be one where there are many entry points because building a product is just necessarily this very Complex, multifaceted thing.”
Yamashita: Code is the primary source of truth in Figma Make
“In Figma make, source of truth is code, and we figure out how to take your design, translate it into code, and then let you iterate via code, because code is so expressive.”
Yamashita: Prompting and coding are as valid as canvas manipulation for design
“Design you might think of as that thing of visual manipulation on a canvas drawing some mock, but actually just as valid is A bunch of iterations via prompting or just as valid is writing some code because maybe designing via coding is faster.”
Yamashita: Designers rarely push pixels today, instead compositing design system elements
“Designers, they're very rarely pushing pixels anymore. They're compositing things that Already exist in their design systems and putting it together.”
Yamashita: Product excellence requires tension between business, user, and tech advocates
“And that tension is what I think advances a product, right? And so I think people have to push on these things, like advocate for the business, advocate for the user, advocate for technical feasibility, and they might not be expressed through these traditional…”
Yamashita: Figma engineering headcount won't scale proportionally with product expansion
“I don't think we'll have fewer engineers than we have today, but you know, I think we'll be able to do more than we're able to today. You know, as we expand more products, I don't, expect the number of engineers, for example, to grow in proportion.”
Yamashita: AI tools lower entry barriers while boosting top-tier productivity
“I always like to go back to our own framework of how AI helps with design, which is that it does both lower the floor and raise the ceiling, right, which is that it makes, in our case, design, but in this case, coding more accessible and approachable to everyo…”
Stebbings: Avoiding video podcasting on YouTube was a major strategic mistake
“A good example for me, say, would be I was very much a fan of not doing video. We don't do YouTube. No, we are constrained on audio because we do a lot of shows, but they're very, very good. People don't feel as comfortable on video. The constraint of channel …”
Yamashita: Non-designers prefer linear workflows over Figma's 2D infinite canvas
“But as we started building other products, we realized, and other audiences, they're like, oh no, people are confused by a two D canvas. They would prefer to be in more of a workflow where they can see one thing at a time. And so as we contemplated other Types…”
Yamashita: Figma tests new feature stickiness with a one-week usage check
“We often do some beta testing where we put it in front of a team and then we check in a week later and are they still using it?”
Yamashita: Figma's early Make Design feature merely matched prompts to templates
“We had worked on some AI features the year before, which at the time we had called make design, where you can go from a prompt and it would generate some designs on the canvas. And it was a pretty simple feature. You write some prompts and then it would use so…”
Yamashita: Figma Make explicitly uses Anthropic's Claude AI model
“Today, like, if you take a screenshot of Figma Make, you see, like, oh, it's being powered by Claude, and it's like, okay, like, no questions. You know exactly where the code is being generated from.”
Yamashita: Early-stage startups should not hire product managers
“Smaller companies shouldn't have PMs because the founder is playing that role.”
Yamashita: Great PMs must frame routine tasks as existential missions
“A PM is really great at motivating something incredibly boring as existential, and so whenever I'm interviewing PMs, you know, I just ask, okay, well, what are some projects you've led, and whatever, and if they can't make it sound interesting to me, no matter…”
Yamashita: Figma's perfectionist culture requires pushing teams to ship
“For example, at Figma, we have a culture of everyone wants to get the details right, so sometimes we have to encourage people to be comfortable getting something shipped.”
Yamashita: Polish core user paths first, skip perfecting speculative features
“If you're perfecting details for a critical part that it's going to help you understand how well this product is going to do in the market, then by all means. But if you're starting to kind of perfect these areas that are much more speculative or areas of the …”
Yamashita: Figma tracks user reliance via ND7s, targeting 5 days weekly
“Our first product, Figma Design, has always set the bar for what it means to be a kind of tool that people are deeply, deeply reliant on. That's kind of the golden standard... We call it ND sevens, like how many days in a week? That people are using your produ…”
Yamashita: Dylan Field mapped design Twitter graphs to recruit early users
“One of the first things he did was he took design Twitter, he visualized it as a graph, and then figured out which people designers are following the most, and then went to those designers to get feedback on Figma and get them excited about Figma. And that was…”
Yamashita: Software Tools Differ From Consumer Apps Due to Unpredictable Behavior
“Some of these are pretty sophisticated products, and the kind of interesting thing about building a tool is a tool is very different from, like, a typical consumer product where, you know, there's a funnel and you're trying to optimize the metric. Like, you're…”
Yamashita: Designers present idealized mocks in portfolios instead of production screenshots
“There's still a strong disconnect between what people design and what ends up in users hands. And for me, a thought experiment is next time a designer shares a portfolio with you, are they sharing the designs that were beautiful mocks? Are they sharing the des…”
Yamashita: Gap between design and production code is primarily a bug
“I would say it's primarily bugged today. Perhaps at times masquerading as a feature because I think designers are always designing ahead, right? So they're always thinking about v one, v two, v next. And so as a result of that, it is natural for the designs to…”
Yamashita: Small 'Little Big Updates' are among Figma's most popular launches
“In fact, some of our most popular launches are these small little things that we brand them as little big updates. Our users love it.”
Yamashita: Adding headcount does not solve product integration challenges
“It's not always the case that throwing more people or resources at a problem makes something go faster, and I say this because we just launched a bunch of new products, and perhaps the hardest problem still is how do we make them connect better, and making the…”
Yamashita: Design and human craft will differentiate software in the AI era
“There's a sea of software, you know, everyone can make some product, and it's really about, well, what is going to help differentiate One product from the next, and we deeply believe it's great design, great craft, and that's not something that, like, you can …”
Yamashita: Shishir Mehrotra excels at intentionally designing organizational culture
“He's such an incredible systems thinker. He just knows how to run an organization, and he obsesses over rituals and basically kind of designs. I think a lot of times, you know, you think about building a product and designing a product, but he also thinks abou…”
Yamashita: Figma's internal culture relies heavily on undocumented tribal knowledge
“It's still a very tribal knowledge. You know, you kind of have to know this random nuance here, which is why, you know, this backstory, which is why we can't do this. And sometimes that prohibits some of the new people from being able to get to a better soluti…”
Yamashita: Waymo delivers best-in-class consumer experience through detailed safety cues
“The consumer product right now that I love is Waymo, but it's more kind of the physical experience of it. I mean, part of it is obviously just how impressive it is in making you feel incredibly safe in something that, like, didn't seem plausible, and all the l…”