Latest / The CTO Podcast with Fexingo: Technical Leadership, Architecture, and Engineering Org / How Figma Prevented Engineering Burnout
Transcript
- Lucas: If you look at the standard playbook for scaling an engineering org, it almost always involves more hiring, more tools, and slightly longer hours until the next major release. But Figma’s approach to their recent infrastructure overhaul was radically different because they treated engineer capacity as a finite resource rather than an infinite well. They stopped trying to optimize for speed alone and started optimizing for sustainability, which sounds like a soft concept until you realize it directly impacted their bottom line. Luna: That is a sharp distinction, especially given how many companies are still pushing four-day work weeks just to keep people from quitting. Did they have a specific trigger event that forced them to change course, or was this a proactive shift? Lucas: It was definitely reactive, triggered by a severe degradation in platform reliability during the third quarter of twenty twenty five. The core rendering engine began experiencing intermittent latency spikes that weren't immediately visible in standard monitoring dashboards. The engineering team responded with what they called 'heroic sprints,' pulling all-nighters to patch the issue, but the fixes introduced new bugs that cascaded into other services. Luna: So the classic vicious cycle where exhaustion leads to mistakes, which leads to more exhaustion. How did they actually measure the cost of that burnout before making the pivot? Lucas: They introduced a metric they called 'context-switching tax,' which quantified the productivity loss caused by interrupting deep work. By analyzing commit logs and calendar overlaps, they found that engineers were spending nearly forty percent of their week in meetings or reacting to alerts rather than writing code. This wasn't just about morale; it was about throughput. Every hour spent in a status update call was an hour not spent debugging the very issues causing the latency. Luna: I love that metric because it makes the invisible visible. Most teams track lines of code or tickets closed, which are terrible proxies for actual value creation. What was the first concrete policy change they implemented to reduce that tax? Lucas: The most surprising move was mandating two full days of no-meeting focus time per week for every single individual contributor, regardless of seniority. This included product managers and designers, not just engineers. They also eliminated all recurring weekly syncs that didn't have a clear decision-making agenda, replacing them with asynchronous updates via shared documents. It felt counterintuitive at first because leadership was worried about losing visibility into project progress. Luna: But they must have had data to show that less visibility through constant check-ins actually led to better outcomes. Did they see immediate results in the code quality? Lucas: Yes, within three months, their average cycle time from commit to production decreased by eighteen percent. Engineers were able to stay in flow states long enough to tackle complex architectural problems without being pulled back into the context of previous tasks. The quality of code reviews also improved significantly because reviewers had dedicated time to read diffs thoroughly rather than skimming them between meetings. Luna: That is a massive improvement in such a short window. I am curious if they faced resistance from middle management who were used to the old way of operating. Usually, those layers feel threatened when you remove the meetings that justify their existence. Lucas: There was definitely pushback initially, particularly from program managers who relied on daily standups to track dependencies. To address this, Figma invested heavily in automated dependency mapping tools that updated in real-time. Instead of asking people what they were doing, the system showed exactly where blockers were forming. This shifted the role of management from tracking activity to removing obstacles, which is actually what they should have been doing all along. Luna: Shifting from policing to enabling. That feels like a fundamental cultural reset rather than just a process tweak. How did they handle the transition for teams that were already burning out? Did they offer any support structures? Lucas: They implemented a 'recovery sprint' protocol where teams affected by the earlier crisis were given a two-week period with zero new feature commitments. Their only goal was to stabilize existing systems and refactor technical debt. This was controversial internally because stakeholders wanted new features shipped, but leadership stood firm. They argued that you cannot build on a foundation that is actively crumbling. Luna: Standing firm against stakeholder pressure is often the hardest part of this kind of transformation. Did the business side eventually accept the trade-off of slower feature velocity for better stability? Lucas: Absolutely. Once the platform latency stabilized and customer churn dropped by twelve percent, the business case became undeniable. The marketing and sales teams saw that reliable performance was a stronger selling point than rapid feature releases for enterprise clients. This alignment allowed engineering to maintain their sustainable pace without feeling like they were holding the company back. Luna: It is interesting that reliability became a competitive advantage. In my experience, most companies treat uptime as a baseline expectation rather than a market differentiator. How did they communicate this shift to the broader organization? Lucas: They published an internal 'engineering health report' that was shared company-wide. It included transparent metrics on developer satisfaction, system reliability scores, and the correlation between focus time and output quality. By making the data public, they created accountability across departments. If product demanded urgent changes, they had to acknowledge the impact on the health metrics. Luna: Transparency as a lever for behavioral change. That is a powerful tool. Do you think this model is scalable for smaller startups that don't have Figma's resources to build custom dependency mapping tools? Lucas: The principles are scalable even if the tools are not. At its core, this is about respecting cognitive load. Smaller teams can achieve similar results by simply protecting focus time and automating routine status updates using basic integrations. The key is leadership willingness to prioritize long-term sustainability over short-term heroics. You do not need fancy software to stop scheduling unnecessary meetings. Luna: Right, the mindset shift comes before the tech stack. I wonder if other industries could learn from this. Healthcare and education seem to suffer from similar administrative bloat. Does the same logic apply there? Lucas: It applies anywhere knowledge work happens. The underlying mechanics of attention and context switching are universal. Whether you are coding a design tool or managing patient records, interruptions degrade performance. The challenge in non-tech sectors is often deeper structural inertia, but the starting point is the same: identify where your experts are wasting time and give them space to do their best work. Luna: That is a really compelling parallel. It suggests that the problem isn't unique to Silicon Valley but is a broader issue with how we structure modern professional life. Before we wrap up, I want to touch on one specific detail about their recovery sprint. Did they define success criteria for that period? Lucas: They defined success purely by system stability and team sentiment. If error rates dropped below a certain threshold and engineer satisfaction scores improved, the sprint was considered a win. There were no feature milestones attached. This clarity helped teams avoid the temptation to sneak new work into the recovery period, which is a common failure mode. Luna: Keeping the scope tight is so important. One final question: looking ahead to the rest of twenty twenty six, do you see this sustainable engineering model becoming the industry standard, or will most companies return to crunch mode? Lucas: I suspect we will see a bifurcation. Companies that adopt this model will retain talent and build more robust products, while those that don't will continue to struggle with high turnover and fragile systems. The market is beginning to reward reliability over raw speed, so the pressure is shifting. It might take a few more years, but the direction seems clear. Luna: A future where building things well matters more than building them fast. That is a hopeful note to end on. If these conversations are useful for what you're building or running, remember that listener support keeps this network independent and ad-free. You can find us at buy me a coffee dot com slash fexingo. Lucas: Thank you for helping us keep the signal to noise ratio high. Your contributions allow us to dig deeper into cases like Figma without relying on corporate sponsors who might want us to soften our analysis. Luna: Exactly. And now that we've covered the human side of engineering, let's get back to the technical specifics. How did they actually refactor the legacy codebase during those recovery weeks without introducing new instability? Lucas: They adopted a strangler fig pattern for the refactoring, gradually replacing old components with new ones while keeping the interface stable. This allowed them to test each module in isolation before integrating it back into the main system, ensuring that the platform remained usable throughout the entire process. Luna: Gradual migration instead of big bang rewrites. A practical lesson for anyone staring down a messy database today. Lucas: Precisely. Sometimes the best architecture is the one that lets you sleep at night.