Skip to content

GitHub reclaims 72,000 engineering hours annually with DX

How GitHub drives $11.4M in annual ROI

GitHub

Challenge: Insufficient data hid developer friction across legacy systems

GitHub, home to more than 150 million developers and the platform of GitHub Copilot, has long been at the center of how software gets built. With thousands of engineers maintaining both modern services and its decades-old monolith, GitHub views developer productivity as critical to its ability to innovate, attract talent, and deliver value to customers.

“Developer friction—or lack thereof—is a powerful signal and vehicle to drive change,” explains Jakub Oleksy, SVP of Engineering at GitHub. “As we think about becoming a better engineering team, it’s really important to understand where we’re falling short so we can make the proper investments to address them.”

“As we think about becoming a better engineering team, it’s really important to understand where we’re falling short so we can make the proper investments to address them.”
Jakub Oleksy, SVP Engineering, GitHub

Before adopting DX, GitHub relied on in-house efforts to measure developer experience. Every step required manual work, from structuring surveys to analyzing results. “Previously, everything was hand-rolled,” says Meirav Feiler, VP of Engineering. “We basically got nothing for free. Even putting the survey out, structuring the questions themselves, and then figuring out how to turn the responses into insights—it was all on us.”

Another challenge was acting on the data. Insights remained high-level, leaving GitHub unable to pinpoint which teams were struggling most. “We knew certain points of friction weren’t company-wide issues. Sometimes they can be very localized challenges,” Feiler adds. “But with our old system, we really struggled to get the insights into the hands of the people who could actually make changes.”

Without reliable data, prioritization often defaulted to anecdote. “Without having focused data, the loudest voice tends to win,” Oleksy explains. “But it might not actually be the highest return on investment.”

 

Solution: Targeted insights for leadership and teams that saved 72,000 engineering hours

GitHub selected DX because of its ability to combine qualitative and quantitative data while keeping developer friction at the center. “DX solves the critical problem of visibility,” says Oleksy. “Previously, we only had high-level views, but the real value comes from specificity. DX puts that data directly in the hands of leaders and teams, so they can see what matters locally and act quickly and effectively.”

“The experience, the polish, the ease of use—it was clear this wasn’t just an executive tool. We could hand it to everybody and they’d get value out of it right away.”
Jakub Oleksy, SVP Engineering, GitHub

Just as important, DX’s product design made adoption effortless across the company. “The experience, the polish, the ease of use—it was clear this wasn’t just an executive tool,” says Oleksy. “We could hand it to everybody and they’d get value out of it right away.”

DX data is available to everyone at GitHub, providing macro-level insights for executive leadership while empowering individual engineering managers to resolve bottlenecks locally. By integrating measurement into daily operations, DX transitioned into an essential decision-making tool for leadership. GitHub now sees a sustained 95% DX participation rate across its developer population. This has enabled leaders to make informed decisions across areas such as their monolith versus non-monolith investments, maintenance costs, CI and build friction, as well as equip managers with insight into key team-level investments they should be focused on.

This transparency has even sparked cultural change. “One thing I kicked off is a friendly competition: which director can improve their DX scores the most by the next quarter? Historically, we’ve said the Platform team has to solve everything, but now we’re empowering teams to act locally,” says Feiler. “The data makes that possible.”

“Historically, we’ve said the Platform team has to solve everything, but now we’re empowering teams to act locally. The data makes that possible.”
Meirav Feiler, VP Engineering, GitHub

DX also helps GitHub measure and improve its AI SDLC tooling, including GitHub Copilot. Eirini Kalliamvakou, Staff Researcher, shares, “The insights from DX enrich our understanding of how AI is impacting developer productivity, and where we can invest to improve AI tooling and adoption.”

These targeted efforts resulted in a 6-point improvement in GitHub’s Developer Experience Index (DXI), and helped reclaim 72,000 engineering hours annually. Moving forward, GitHub plans to leverage DX to keep a pulse on how their investments in improving the delivery process are paying off.