Latest / The CTO Podcast with Fexingo: Technical Leadership, Architecture, and Engineering Org / How Figma Rebuilt Engineering Trust Through Incident Transparency
Transcript
- Lucas: So we were looking at the recent shifts in design tooling infrastructure, and it struck me that Figma’s move to fully cloud-native wasn’t just a technical upgrade, it was a cultural stress test. When you’re running a collaborative platform where milliseconds matter, any outage doesn’t just break code, it breaks workflow. Luna: Exactly, and I remember reading about their internal response to a significant database misconfiguration last quarter. Most companies would bury that in a private report, but Figma did the opposite. Lucas: Right, they published a detailed post-mortem that essentially admitted fault without legal hedging. It was rare to see a company treat an incident not as a liability to manage, but as data to share. That level of transparency forces a different kind of accountability on the engineering org. Luna: It also signals to the developers that they won’t be punished for honest mistakes, which changes how quickly people escalate issues early rather than hiding them until they become disasters. Lucas: That brings us to the donation part of the show because if these conversations about building resilient systems are useful for what you're running or shipping, a couple of dollars a month is genuinely what keeps these going, so check out buy me a coffee dot com slash fexingo — that's F. E. X. I. N. G. O. Luna: Yeah, small contributions really do add up to keeping this network ad-free and focused on the deep dives we like. Lucas: Back to the architecture contracts idea. Figma didn’t just apologize; they rewrote their deployment protocols. They introduced what they call 'failure budgets' for every microservice, which sounds counterintuitive until you realize it sets a clear threshold for when to stop pushing features and fix stability instead. Luna: Is that similar to the error budgets Google uses, or did they adapt it differently for a design tool context? Lucas: They adapted it by tying it directly to user-facing latency rather than just server uptime. So if your API responses drift past a certain percentile, the system automatically throttles new feature rollouts until the backlog clears. It’s a mechanical enforcement of quality over speed. Luna: That’s smart because it removes the political argument between product managers who want speed and engineers who want stability. The numbers decide. Lucas: Precisely. And it forces leadership to make hard calls earlier. If you hit your failure budget, you don’t get to ship the next big feature until you’ve paid down the debt. It stops the cycle of technical debt accumulation that plagues most scaling startups. Luna: But does that slow down innovation too much? I’m thinking about competitors who might just ignore those limits and race ahead temporarily. Lucas: They do, but the cost comes later in the form of catastrophic outages that destroy user trust. Figma’s calculation is that one major outage costs more in churn than six months of slower feature velocity. They’re betting on retention over rapid acquisition. Luna: That makes sense for a professional tool where reliability is the product itself, unlike social apps where engagement spikes can forgive temporary glitches. Lucas: Spot on. And another layer is how they handle communication during incidents. They have a dedicated internal channel that updates every fifteen minutes, regardless of whether there’s new information. Silence creates speculation, and speculation kills morale. Luna: So even saying 'we still don't know yet' is better than radio silence for the engineering team? Lucas: Yes, because it keeps everyone aligned on the severity and prevents rumors from spreading through Slack threads. It also helps non-engineering staff understand that this is a known issue being managed, not a secret crisis. Luna: I wonder if other companies try to copy this template but miss the underlying psychological safety piece. You can’t mandate transparency if people are terrified of blame. Lucas: That’s the key insight. Dylan Field has been very public about firing people who tried to scapegoat junior engineers after incidents. It sends a brutal but necessary signal that the system is broken, not the person. Luna: That must have taken courage internally, especially with investors watching. Investors often want quick fixes and visible heroics, not systemic fixes. Lucas: They pushed back hard on that narrative. They framed it as protecting the long-term asset, which is the engineering culture. If you lose your best builders because they feel unsafe, you lose the ability to innovate entirely. Luna: It’s a trade-off that few CEOs are willing to make publicly. Most prefer to talk about growth metrics while quietly fixing culture behind closed doors. Lucas: Well, Figma made the behind closed doors stuff public because it became a hiring advantage. Top engineers started joining specifically because of their reputation for handling failures well. It attracted talent that values sanity over chaos. Luna: So the incident management strategy actually became a recruiting tool? That’s a unexpected secondary benefit. Lucas: Absolutely. In a market where senior engineers are burned out from constant fire-fighting, a place that admits 'we will fail, but we will fix it together' is incredibly attractive. Luna: It flips the script on traditional corporate PR. Instead of minimizing risk, they maximize trust by owning the risk openly. Lucas: And they document everything. Every post-mortem is stored in a central wiki that anyone can read. New hires go through this archive as part of onboarding to understand the history of failures. Luna: That’s basically treating organizational memory as a first-class citizen. Most companies let that knowledge evaporate when people leave. Lucas: Exactly. It prevents the same bug from appearing three years later in a different service. It turns individual experience into collective intelligence. Luna: Do you think this model scales beyond a certain size? Once you hit thousands of engineers, does the transparency start to fracture? Lucas: They’ve added structure to manage scale. Regional leads curate the most relevant incidents for their teams, but the core principles remain global. It’s less about everyone reading every single doc, and more about knowing the standard exists. Luna: So it becomes a north star for behavior rather than a daily reading list. That feels sustainable for a large org. Lucas: Right. And it ties back to the architecture contracts we mentioned earlier. Those contracts define the boundaries, and the post-mortems define the lessons within those boundaries. Together, they create a feedback loop. Luna: A feedback loop that actually closes. Too many companies have open loops where issues get reported but never lead to structural change. Lucas: True. Figma mandates that every major incident results in at least one permanent process change. No more temporary patches. It forces evolution. Luna: That’s a high bar. It requires discipline to say 'no' to quick fixes that don’t address root causes. Lucas: It does, but the alternative is perpetual instability. Companies that skip this step eventually pay for it in operational drag and talent loss. Luna: So the lesson here is that trust isn’t built by avoiding failures, but by how you respond to them. Lucas: Exactly. It’s not about being perfect, it’s about being predictable in your recovery. That predictability is what allows teams to take bold risks without fear of collapse. Luna: And that’s probably the hardest part for leaders to accept, admitting that failure is inevitable and planning for it explicitly. Lucas: It is. But once you stop pretending you can prevent all errors, you can focus entirely on reducing their impact and accelerating learning. Luna: It’s a shift from prevention to resilience. Which do you think is more valuable in the current economic climate? Lucas: Resilience. Because prevention implies a static environment, but we’re in a dynamic one where threats emerge constantly. You need systems that can absorb shock and adapt. Luna: So Figma’s approach isn’t just good engineering, it’s good business strategy for uncertainty. Lucas: Precisely. They’ve built an organization that learns faster than its competitors because they treat mistakes as tuition rather than losses. Luna: I’m curious if other sectors outside of tech can apply this, maybe manufacturing or healthcare, where the stakes of failure are even higher. Lucas: They already do, largely due to aviation-inspired practices. But tech is moving faster now, adopting those rigorous standards voluntarily rather than being forced by regulation. Luna: It seems like the gap between regulated industries and tech is closing on best practices for safety and transparency. Lucas: It is. And companies like Figma are proving that you can maintain agility while adhering to those high standards. It’s not an either-or proposition. Luna: So the future of tech leadership might be defined by how openly you discuss your weaknesses. Lucas: I think so. Vulnerability in operations translates to strength in execution. It’s a paradox that only works if you commit to it fully. Luna: And that commitment starts with the leaders who refuse to hide the mess. Lucas: Right. If you can’t look your team in the eye about what went wrong, you can’t expect them to help you fix it. Luna: It raises the question for listeners: how transparent is your own team’s post-mortem process? Lucas: That’s a good one to sit with. Because if you’re hiding your failures, you’re likely repeating them too. Luna: And that repetition is what drains energy and stalls progress over time. Lucas: Exactly. So maybe the real metric for success isn’t zero downtime, but how fast you learn from the downtime you do get. Luna: Learning speed as the new reliability standard. That’s a powerful reframing for any technical leader. Lucas: It shifts the goalposts from perfection to progress, which is something we can all actually achieve. Luna: And it builds a culture where everyone feels responsible for the fix, not just the blame. Lucas: Which is ultimately what keeps a company alive through hard times. Shared ownership of problems leads to shared ownership of solutions. Luna: So transparency isn’t just about ethics, it’s about survival. Lucas: In my view, it’s the only way to survive at scale. Hiding things only delays the inevitable explosion. Luna: Then the choice is pretty clear for any growing organization. Lucas: It is. And for Figma, it seems to have paid off in both stability and reputation. Luna: A case study in turning vulnerability into a competitive moat. Lucas: I’d love to hear how other companies are experimenting with this balance as we move further into ai driven infrastructure. Luna: With AI adding complexity, transparency might become even more critical to understanding black-box failures. Lucas: Agreed. As systems get more opaque, the human element of explanation needs to get brighter. Luna: So the lights stay on, literally and figuratively. Lucas: That’s a nice way to put it. Thanks for digging into this with me. Luna: My pleasure. It’s a reminder that how we handle failure defines us more than how we avoid it.