Introducing Span's AI Effectiveness suite, powered by agent traces
Air
How Air built a data-driven leadership practice around engineering observability
250+ employees
Creative Operations
4
mins read
55% → 65%
Allocation to new features within one planning cycle
Review Bottlenecks
Rebalanced across roles and levels
“Every engineering organization is different. Span lets me observe mine — where we're really strong, and where we have room for improvement — and make decisions based on real-world data instead of gut feeling.”
Ping Ma
VP of Engineering
Air is a creative operations platform that helps teams manage their creative assets, no matter the scale. Ping Ma joined as VP of Engineering and leads a 25-person team across frontend, backend, and platform.
He frequently works with leadership at Air to determine engineering priorities, especially as Air continues to lean more heavily into new AI features for their product.
Air's leadership needed a shared data surface to help steer engineering decisions
Before Span, Air's engineering team was already invested in industry frameworks like DORA and SPACE, and had piped GitHub metrics into Mode to track PR lifecycle, time-to-review, and weekly deploy counts. But the dashboards were scattered, and reviewing them had not yet become a consistent part of how the team refined its workflows.
Additionally, the company was preparing a major strategic shift in H2 2025 that required refocusing on AI feature development, but there was no unified way to monitor whether the shift was actually happening. This created challenges during cross-functional conversations when Air’s leadership and Ping were trying to align on engineering priorities, as they needed more concrete data that would give them the visibility they needed to make decisions.
Span became the observability layer for Air's engineering organization
After adopting Span, Ping and his leadership team gained deeper visibility into the development lifecycle from multiple angles — productivity, code review patterns, individual team health, and how engineering time was split between feature work, KTLO, and enhancements. Investment Mix gave leadership a view of how effort was actually distributed across the roadmap, while PR-level metrics surfaced patterns in code review that had previously been difficult to see clearly.
Span connected directly to the team's existing GitHub workflow and replaced the scattered Mode dashboards with a single view leadership could return to on a regular basis.
A leadership practice built on shared data
Engineering data brought new depth to leadership reviews
Ping meets with Air’s leadership team on a bi-weekly basis to discuss where engineering data sits alongside product metrics. The cadence predated Span, but the level of insight Ping can bring into the room has changed. He pulls from Span — weighted PRs per week, PR comment volume, time from first commit to first review — and gives leadership the context they need as he walks through the data.
With that visibility, Ping and the Air’s leadership have been able to more clearly align on priorities and drive engineering decisions based on data-driven evidence.
“In the past we would talk about this kind of thing conceptually, but without visibility into the data we couldn't really align as a leadership team,” said Ping. “Now we can talk about it transparently.”
A strategic shift toward AI features, tracked in real time
Heading into 2025, Air's leadership made the call to lower allocation to KTLO and maintenance, and put more headcount behind new AI feature development. The goal was to evolve Air’s product from a system of record into a system of intelligence.
Span was the monitor that helped track whether that allocation was occurring. Ping could watch Air's new feature allocation move from roughly 55% at the start of 2025 to around 65% by the end of 2025, giving the wider organization confidence that the strategic shift was landing in the work the team was actually shipping. Gathering that kind of directional read would have been significantly harder without a single dashboard to return to.
Code review coverage came into focus across roles and levels
One of the patterns Span helped Ping identify was that code review was concentrated among a handful of senior engineers who had built up deep institutional knowledge over the years. On a lean team of 25, that kind of concentration creates real bottleneck risk and limits how the team can scale review capacity as the organization grows.
With that visibility, Ping uses Span as an ongoing lens to make sure review responsibility is distributed accordingly. Staff engineers are expected to contribute meaningfully to code review, and Span gives Ping a clear read on whether that expectation is being met.
Looking Ahead
Air continues to use Span as the observability layer for its engineering organization, with leadership reviews anchored to the data on a bi-weekly cadence.
As the team scales and the AI roadmap accelerates, Ping expects Span to remain the lens through which leadership measures whether the organization is investing where it intends to.
Everything you need to unlock engineering excellence