Most engineering leaders know they should be measuring more. What holds many of them back is a quieter concern about whether the organization is actually ready to see the numbers.
This tension, however, did not keep Ganesh Datta, our co-founder and CTO, and Randy Shoup, SVP of Engineering at Thrive Market, from diving down this rabbit hole on the Braintrust podcast. After decades of leading engineering teams at Google, eBay, Stitch Fix, and now Thrive Market, Randy says that no matter where you are, making things visible is how you start improving them.
That sounds straightforward and borderline obvious, of course, but the nuts and bolts of making this a reality are usually more complicated. In fact, Randy says there are a few components to visibility that tend to matter far more than most leaders expect.
Why some organizations actively avoid transparency
Randy says he looked to the work of Ron Westrum to understand why measuring things makes some teams uneasy. Westrum, a sociologist whose framework for organizational culture appears throughout DevOps literature, says he looks at three types of cultures:
Pathological cultures run on fear. The messenger gets shot, new ideas get crushed, and people protect themselves by keeping what's actually happening out of view.
Bureaucratic cultures are rule-bound and rigid, grinding out middling engineering and business results in roughly equal measure.
Generative cultures are the ones worth building toward. When something goes wrong, the response is inquiry and learning rather than blame.
This framework, Randy says, explains why engineering leaders get so much pushback on measuring things. At eBay, he adds, this wasn't the result of laziness or bad intentions. Rather, it was the result of a fear-based environment in which people protected themselves by being risk-averse and opposed to transparency. It was also entirely predictable behavior of people who had learned that being transparent about problems made their situation worse, not better.
"The only way to get better is to be very open and honest and straightforward with yourself and others about where you currently are." Randy Shoup — SVP of Engineering, Thrive Market
Pathological cultures don't just feel bad to work in, but they consistently produce the worst engineering and business outcomes of the three types, and the reason is information flow. When fear governs how people communicate, problems get hidden rather than surfaced, and problems that don't get surfaced don't get fixed.
Resistance to transparency isn't exclusive to fear-based organizations
Even organizations that don't operate out of fear can run into trouble when they start measuring things out in the open. At one point, Randy and his team at eBay encountered two distinct versions of this anxiety, and learning to navigate both turned out to be just as important as the metrics work itself.
In a pathological organization, teams resist transparency because they genuinely fear what exposure will mean for them personally. Shining a light on current performance feels threatening rather than useful. In a healthier organization, Randy says that leaders start to worry about how measuring things in public might be misread as surveillance, even when the intent is entirely constructive. That worry alone, he says, can be enough to slow everything down.
Most leaders expect pushback from struggling teams, but fewer expect it from good ones. At eBay, Randy says engineers with good intentions still worried that the data would eventually be used against them. Naming both fears out loud, and being explicit about why the metrics existed, was part of how Randy and his team started to build the trust they needed to move forward at all. "The best way to demonstrate trust is to start by being vulnerable a little bit," he continues. "We are introducing these metrics, here's why we're doing it, and we completely understand that it's going to freak you out."
Trust is the foundation, and it builds through behavior over time
When Randy returned to eBay in 2020 as chief architect, he was given the task of speeding up development and engineering, which were too slow relative to their industry peers. He and his partner Mark Weinberg used the DORA metrics as the framework for measuring software delivery, and then worked with a set of pilot teams to start making progress.
The first challenge was less technical and more psychological. Randy's approach was to be direct. He told teams what metrics were coming, explained the reasoning behind them, and openly acknowledged upfront that the whole thing was going to feel uncomfortable. Naming the anxiety rather than hoping it wouldn't surface turned out to be one of the most effective moves they made. Trust doesn't arrive by announcement. It builds through consistent behavior over time, and the behavior that matters most is using the data exactly the way you said you would.
The unit of measurement is the team, not the individual
Randy has one really firm boundary that he says is often blurry in many engineering organizations. He believes metrics like deployment frequency, change failure rate, lead time, and MTTR are measures of how a system is performing, but not indicative of how an individual engineer is doing their job. Applying them at the individual level tends to produce the culture of fear that makes transparency impossible to build in the first place.
"The reason why it takes 10 is the system, not the individual." Randy Shoup — SVP of Engineering, Thrive Market
When delivery is slow, Randy says, the cause is almost always in the system. No engineer drags a one-hour task to ten hours on purpose. The correct unit is the team, and that's the only level at which these metrics tell you something worth acting on.
At eBay, Randy took this principle further than most organizations are willing to go. Every team could see every other team's metrics, which was something nobody else in the organization had even considered previously.
The key was framing, Randy says. The goal was never to compare Team A to Team B. Pointing out the gap between a high-performing team and a struggling one, and using that gap to pass judgment, is a reliable way to destroy the trust you've spent months building. The more useful comparison, as Dr. Nicole Forsgren and Abby Kearns argue in Frictionless, is to compare a team to its own past performance. Showing a team they're better than they were last month generates motivation in a way that ranking teams against each other never does.
Turning shared visibility into shared learning
To make full visibility work in practice, Randy ran a weekly team of teams meeting. He brought together at least one member from each of the pilot teams for an hour every week, and the tone was deliberately celebratory.
Teams shared what they'd tried, what had moved their numbers, and what hadn't worked. When a team was struggling, the room's response was to offer help, not render a verdict. Randy's working assumption was that developers are craftspeople who want to do good work, and that slowness almost always reflects a system problem rather than a motivation problem. When a team is visibly stuck, the right question is what help looks like tomorrow, not who is to blame.
Visibility does most of the cultural work on its own
There's a version of building a measurement culture that runs on mandates and top-down pressure. It can work in the short term, but it tends to reproduce the same fear-based dynamics that made transparency difficult to introduce in the first place.
Randy says transparency itself does about 80 percent of the cultural work. We tend to see this most often when we look at cloud expenses. Show engineers what their infrastructure actually costs, and many of them will immediately start looking for ways to bring it down. Not because they're under pressure to, but because engineers are naturally motivated to solve problems they can actually see.
Celebrating teams that improve is the other part of the equation. Sustainable cultures run on positive reinforcement, and recognizing progress publicly is a more durable driver than consequences for falling short.
The first move is always the same
Across every context Randy described, whether delivery performance, reliability, cloud costs, or production readiness, the starting point was to focus on your people. If you want people to care about a thing, start measuring it and make the measurement visible.
Transparency alone won't close the gap. The framing matters, trust takes time to build, and the rituals you create around the data shape whether visibility becomes a source of motivation or a source of anxiety. But none of the rest of it is accessible until the organization is willing to look at its current state honestly and share what it finds.
Want to listen to Ganesh and Randy's full conversation? Check out the Braintrust podcast here.
