Figma's UX Debt Pixel Audit

Grant: Toot toot Maya: Welcome back to UX Autopsy. I'm here with Maya and today we are doing a full dissection of Figma heading into Config 2026. Speaker 3: June 23rd, Moscone Center, San Francisco. In-person tickets are already sold out according to the Config site, but you can still watch virtually for free. Maya: Which honestly might be the smarter move if you want to take notes without being jostled. Speaker 3: Fair. So here's what we're covering. Today we start with Figma's final... financial story because the numbers are wild. Maya: Yeah, Q1 revenue hit $333 million, up 46% year over year. Net dollar retention at 139%. Stock is still sitting around $22 after peaking at almost $142 last August. Speaker 3: Wait, 142 to 22? That's not a correction, that's a cliff. Maya: Right? Same company, two completely different stories depending on which line. Which line of the financials you're reading. Speaker 3: Um, no. So we dig into that, and then we get into the actual product. Maya takes a hard look at the Figma variables panel. Maya: And I'll be the one defending it, which I am only partially prepared for. Speaker 3: You know what I mean. It's a power tool with some sharp edges. Maya: Then we hit Dev Mode and Figma Make. Here's the thing. There's a big gap between the design to code marketing at... Figma and what the output actually looks like in production. Speaker 3: And we close with a quick fire red flag or green flag on the new AI features. Then one concrete thing you can steal before the keynote drops. Maya: All right, let's get into the money first. Figma's IPO arc is something. Speaker 3: Oh, it really is. Let's go. Maya: Okay, so picture this. Figma prices its IPO at $33 a share in July 2025. Day one, it closes at $115. That's a 250% pop. By August, it's at $142. Speaker 3: Wow. Maya: Then the stock spends the next several months essentially falling off a cliff. Speaker 3: And that's the backdrop walking into Config 2026, which, according to Figma's own site, lands June 20th. 23rd through the 25th at Moscone Center in San Francisco. Maya: Right. So this is the first proper Config since Figma went public and the stock is trading around $22 right now, well off that peak. The market is watching. Speaker 3: I mean, I get why you'd frame it as a reckoning, but Grant, look at the actual business. Figma just reported Q1 revenue at $333 million, up 46% year over year. They beat their own guidance. They raised the full year outlook. Maya: Wait, 46%? Yeah, I know, the fundamentals are genuinely solid, but here's the thing. The stock tells a different story than the income statement. The market isn't pricing a company growing at 46%. It's pricing a company that has to prove something. Speaker 3: And what does it have to prove specifically? Maya: That it can own the AI era on its own terms. Because remember, Remember, the Adobe deal, that collapsed in December 2023. Regulators in the UK and EU killed it, so Figma had to build its whole AI stack independently. Meanwhile, Adobe has been pouring around $3.8 billion into AI. Figma can't match that spend. Speaker 3: Okay, fair, their resource gap is real, but the product is clearly resonating. Net dollar retention hit 139% in Q1. That's customer spending more, not less. Maya: No, I know. The tool is sticky. That's not the question. The question is whether Config lands as a product vision statement or just a feature parade. Speaker 3: And that's where the speaker lineup is interesting. Figma's blog put out a piece on the Config 2026 speakers, and the framing is very specifically around AI and craft, not just AI and speed. They're asking, what do product builders do when generation gets cheap? cheap. Maya: Which, you know, is either a genuinely interesting design question, or it's great marketing for a company that needs the design community to feel like Figma is still the center of the universe. Speaker 3: Could be both. Maya: Totally could. UiuxTrend notes the conference has over 75 design leaders and 50 plus sessions. Virtual attendance is free, so the reach is real. Speaker 3: Here's the thing, though: for anyone watching the keynote, the pressure question isn't Doesn't what features are they shipping, it's whether the product itself holds up under the weight of how teams actually use it. Maya: Exactly, because there is a gap and it's been widening and it shows up in a very specific part of the product that basically every professional team is living inside right now. Speaker 3: So what specifically breaks when you stress test it? Maya: Yeah, so while we're inside the file, the variables panel, because this is where professional teams are actually living right now. Speaker 3: Yeah, and the numbers back that up. According to a Supernova survey, 69.8% of teams have already migrated to Figma variables as their token source of truth. Maya: Right. So this isn't a niche power user thing. It's the daily workflow for most professional teams, which makes the UX problems inside that panel not edge cases. They're everyone's problems. Okay, I'll bite. Walk me through it. Speaker 3: Alright, so you open the variables panel, you've got your collection sidebar on the left, you start building out your token hierarchy, primitive layer, semantic layer, component layer, and Figma gives you basically nothing to work with. No guidance, no validation, no contextual help. Maya: I mean, that's kind of a power user tool, though. Is it reasonable to expect handholding? Speaker 3: Grant, it's not hand-holding I want, it's basic guardrails. You can build a completely broken variable structure, circular aliases, type mismatches, semantic tokens pointing at nothing, and Figma will say not a word. You find out when a component starts behaving wrong in production. Maya: Okay, that's... Yeah, that's a real problem. Speaker 3: And then there's the scope dropdown, hidden at the bottom of the edit panel. Most people never find it. Maya: Wait, the one that restricts which property types the variable can apply to? Speaker 3: That one. It's actually useful, but it's buried so deep that teams just skip it. So you end up with color tokens appearing in spacing fields, number tokens showing up where they shouldn't. The tool has the right feature set. The discoverability is just broken. Maya: Okay, but I'd argue some of this is the flexibility trade-off. The systems Grant: Supports four primitive types: number, color, string, boolean, plus modes, conditional logic, aliasing. That's a lot of power. At some point, complexity is the price of admission. Maya: Sure, I'd buy that for the underlying architecture, but the panel UI doesn't have to be complex just because the system is. Those are two different things. Grant: Fair. Maya: And the export situation is wild. Figma still has no native variable export. export. Like, none. Teams that need to get variables into SCSS or JSON are relying on third-party plugins. One team I read about literally built a custom Claude plugin just to handle their SCSS export. Grant: Wait, really? They built their own AI plug-in to export tokens? Maya: Yeah, because the official path doesn't exist, and this gap has been sitting there for years while the variable system itself kept getting more complex. Grant: I mean, that's wild. Though I'd note the system's complexity is also what makes it worth building custom tooling around. Nobody builds Claude plug-ins for features they don't care about. Maya: That's a very generous read. Grant: I'm a generous guy. Maya: Look, the honest answer is the variable system is genuinely powerful and genuinely frustrating, and I don't think those cancel each other out. Grant: No, they don't, and I don't think Figma would say they do either. The question is whether Config 2026 addresses any of this directly. Maya: Which brings up the next problem, because the designer finishes the file, hits publish, and then a developer opens it, and what they see in Dev Mode is a whole different category of broken. Grant: Oh, the inspect panel. Yeah, that one's rough. So the Designer hits Ready for Dev on the status toggle, the Developer opens the file and... what? An inspect panel that shows you per element spacing and variable names but zero view of the whole token system? You see, color slash brand slash primary on a button. Congrats! You have no idea where that lives in the overall structure. Maya: Right. It's like handing someone a street address, but no map of the city. Grant: And that's the core Dev Mode problem. Figma keeps selling this as a handoff solution, but the inspect panel was built for an element-level question, not a system-level one. A developer trying to understand the full token hierarchy has to go find it somewhere else. Maya: I'll give them the Git integration, though. Live code sync, version diffing built-in. That's actual infrastructure. That's not nothing. Grant: No, that's real. But then you hit the pricing math, and it falls apart. According to multiple sources tracking Figma's seat model, dev seats run roughly $12 to $35 a month, depending on your plan. A full designer seat on Professional is $15. So you're paying close to the same rate for a developer who can only eat. Only inspect. Maya: Wait, and most product teams are running way more developers than designers. Grant: Right? On a typical squad, you might have one designer and five engineers. You want all five in Dev Mode? You're basically doubling your Figma bill. Most teams just don't do it, which means the handoff experience Figma is pitching at Config is something a lot of companies can't actually afford to deploy. Maya: Okay, and two-thirds of Figma's monthly users aren't... aren't designers. A lot of those are developers. So Dev Mode is the second most used workflow surface in the whole product, and it's priced like an afterthought. Grant: That's wild to me. And then there's Figma Make. You know, the design-to-code story they're going heavy on at Config this year. Maya: Oh, here we go. Chuckling. Grant: So I have. According to a UX Collective review by Allie Paschal, Figma Make's code output is, quote, Messy and uses generic elements like div. No semantic HTML. The reviewer literally said exported code has to be restructured and debugged, making it more costly than starting from scratch. Maya: Okay, but that review is specifically about production handoff. Figma's been pretty clear that make us for concept validation and user testing right now. They're not pretending it ships to production yet. Grant: Are they clear? Because the Config marketing frames it as design the code. Oh, Figma's own solutions page says it outputs code that's ready to test, share, or refine. That's not "this is a prototype tool. Maya: I mean, that's fair. The messaging is doing a lot of heavy lifting, and when you're running a keynote, concept validation doesn't exactly move the crowd. Grant: Exactly. And look, I get the long game. The MCP server integration is genuinely interesting. VS Code access to Dev Mode, live design context into your coding agent, that's a real direction. Maya: Right, and the Inspect panel problem is the one developers hit. Hit daily, the MCP route is at least a path around it. Instead of reading the panel, your agent reads the file. Grant: Which brings us to the AI features, because the agent stuff and what they're demoing at Config this June, that's where some of these handoff gaps either get patched or get worse, depending on the pricing. Maya: Nodding. And the credits question is a whole other thing. We should get into that. Grant: OK, so let's do this fast. Red flag or green flag, Figma's AI features. Grant, I'll go first. Maya: Jumping right in. Grant: Red flag, Figma's own release notes confirm credit supplied at general availability for the agent, and the credit system went live for full seat users back in March. So everything you're watching them demo at Config, you will eventually pay extra for it, on top of a bill that's already gone up. Speaker 3: I know, I know, but hear me out before you fully write it off. Grant: Okay. Speaker 3: Green flag on the agent itself. According to Figma's release notes, it launched in beta May 20th, lives directly on the canvas, and actually reads your component library, not a bolt-on sidebar. Grant: That part I actually believe. The MCP integration is real infrastructure. Speaker 3: Right. And it can run bulk edits, swap components across a whole frame. whole file, rename variables for consistency, the stuff that eats an hour of your day. Grant: That's a genuine green flag. The question is what happens to that price tag at GA. Speaker 3: That's the tension, yeah. Now, green flag on Figma Weave. They acquired Weavy back in October, rebranded it Figma Weave. TechCrunch covered it. Image, video, animation, VFX, all on a node-based canvas. Grant: The workflow problem it solves is real. Designers were context switching to Runway, Midjourney, three other tools just to pull assets. Speaker 3: Exactly! One canvas instead of five browser tabs—that's a workflow win. Grant: But here's my flag on it: the Figma blog says the full integration is coming later this year. It's still a separate product with separate pricing. Speaker 3: Okay. Grant: Yeah, that's fair. It's a future green flag, not a today green flag. Maya: Which brings me back to the credits thing. Figma Help Center also notes Weavy credits are managed separately and don't factor into your main AI credit balance. Grant: Wait, so that's a second meter running? Maya: Potentially, and that is a red flag full stop. Grant: On both sides a little bit. The capabilities are real; the pricing architecture is not your friend. Maya: So your Config keynote filter right there asks, is this actually shipping or is this a credit line item I haven't seen yet? Grant: That question will save you a lot of budget conversations in September. Maya: So, headed into the close, all the features we've been picking apart, the agent, the token system, the handoff gaps. Config is the moment Figma either shows a credible roadmap or buys more time on hype. Grant: And whether the price is worth it for your specific team, that's the only answer that matters. Maya: Ah. Grant: So the question on the table for June 23rd is what does Figma actually need to show for Config to matter? Maya: Same way I'd size up any pitch. What's the gap between what you're promising and what you can ship today? Because right now that gap is real. Grant: And the competitive pressure isn't abstract. Adobe Firefly has generated over 24 billion assets, according to SQ Magazine's data. Figma's platform is deeply embedded across Creative Cloud already. Figma Make is sitting at roughly 2% of the AI design tool space. That's not a close race. Maya: No way. And look, Figma's speaker lineup is actually telling. According to Figma's own blog, Matthew Strøm-Awn is on the bill. He's an editor for the W3C Design Tokens Community Group. That's not a random pick. Grant: Right, because the token-to-code pipeline is exactly the gap we've been picking at all episode. If he's on that stage, you'd hope it means Figma is finally going to address native export against the W3C spec, not just a demo that disappears into a plug-in. Maya: Not just a demo should be the whole Config theme at this point. Grant: Honestly. Maya: So here's the practical thing I'd tell anyone watching the keynote. Before you open the stream, write down the three workflows in your current Figma setup that cost you the most time every week. Specific ones, not vague stuff. Grant: Ooh, I like that. Maya: Then judge every single announcement against those three things only. Because keynotes are built to generate excitement. They are not built to solve your actual problems. Grant: And the gap between those two is exactly where hype lives. Like, the canvas agent beta is genuinely interesting, but does it fix the thing that's costing your team three hours on Thursday? Maybe, maybe not. Maya: Ugh. And that's the honest answer on Figma right now. The fundamentals are solid, the AI direction is at least pointed correctly, but can they close the token pipeline, clean up the Dev Mode story, and prove Make produces real code? before competitors just eat the opportunity. Grant: I genuinely don't know, and I think that uncertainty is the right place to land. Config could be the moment they show a credible path, or it buys them another year of coming soon. Maya: All right, that's a wrap on our Figma audit. And honestly, this one got into some real territory. Grant: Yeah, the broken variable structure moment for me. The idea that Figma will sit there quietly while you build a circular alias mess and you only find out when something breaks in production? That's not a niche edge case. Maya: Totally. And look, the stock story versus the revenue story, those are two genuinely different companies depending on which number you read. You're reading. Grant: Right. The market wants proof, not just growth. Maya: So, Config is June twenty third through the twenty fifth at Moscone Center, San Francisco. Virtual registration is still free if you can't make it in-person. Grant: Do that pre-keynote audit exercise we talked about. Seriously, it'll change how you watch the announcements. Maya: And if today's episode made you look at your own designs differently, leave us a review. It actually helps. Grant: Screenshot a UI that bugs you and tag us. We want to see the crimes. Maya: Subscribe so you don't miss next week. Thanks for being here. Grant: See you on the other side of the keynote.