By Natalie Johnson
Programs that run into the billions rarely collapse in a single bad decision. They erode, slowly, at the handoffs nobody is watching: the gap between engineering and field operations, between one shift and the next, between a vendor’s invoice and the deliverable it supposedly covers. Timothy A. Holden has spent his career inside programs at that scale, and his read on failure is unsentimental. The technology is rarely the problem. The problem is that accountability gets diffused until it belongs to no one, and precision quietly leaks out through small gaps that compound faster than anyone expects. His argument cuts against how most large organizations operate, where oversight is a quarterly ritual and ownership is assigned to teams rather than people. Holden treats execution as a daily discipline that lives with individuals, not committees.
Ownership Belongs To A Person, Not A Department
Start with the simplest and most frequently violated rule in program management. “Every deliverable, risk, and open issue has one name on it. Not a team, not a department. A person,” Holden says. “When something has five owners, it has none.” Most organizations know this in principle and abandon it in practice, because distributed ownership feels collaborative and spares anyone the discomfort of carrying a result alone. The cost shows up later, when a risk nobody owned becomes a problem nobody saw coming.
Ownership alone is not enough without a rhythm that makes work visible while it still can be changed. Holden ties goals from yearly to quarterly to weekly to daily, so the connection between today’s task and the larger number is legible to everyone on the program. “If you only measure lagging results, you find out about problems after they’ve already cost you,” he says. “Leading indicators let you correct while there’s still time.” The same standard runs through vendor relationships, where he insists that every invoice line can be traced back to a specific customer, project, site, and deliverable. If the spend cannot be traced, it cannot be managed. And the discipline applies upward too. Holden describes his own role as holding people accountable while making sure they have what they need to succeed, pairing high empowerment with high accountability. He learned the limits of the alternative early, when he tried to hold everything himself and fell behind on all of it. The shift came when his team leads began creating, sharing, and tracking their own action items after every meeting. He stopped chasing people and started clearing obstacles.
Precision Fails At The Seams, Not The Summit
Leaders tend to look for failure in the big calls. Holden looks at the joints. One failure point is at handoffs, where critical knowledge may live in only one or two people’s heads. Each gap looks trivial in isolation. The second common failure is undocumented change: someone makes a reasonable adjustment, nobody else hears about it, and three weeks later a team is troubleshooting a condition no one can explain. Both failures share a trait that makes them dangerous at scale. They are invisible until they are expensive.
The correction matters as much as the diagnosis, because the standard fix is often worse than the defect. Heavy governance imposed mid-program kills the momentum the program depends on. Holden’s answer is the smallest intervention that solves the problem. At one company he consults with, the team did not deploy a heavy change management system. They introduced a minimum viable change process with simple post-verification checklists after major maintenance windows, light enough to adopt in a week and sufficient to close the gap. Three rules govern his corrections: name the problem openly and early, fix the process rather than the person, and keep the change small enough that the team can absorb it while still delivering. “Momentum comes from confidence,” he says, “and people stay confident when corrections feel manageable instead of punitive.”
Automation Can Execute, But It Cannot Answer For The Outcome
As AI moves out of pilots and into live telecom operations, the accountability question gets sharper rather than easier. Holden’s position is blunt. “An AI system can execute a task, but it can’t be accountable for it. Every automated decision still needs a human owner.” He frames the work in terms of decision rights, defined before anything goes live: which decisions the system recommends, which it carries out on its own, and which always require a person. That framework, he argues, matters more than the tool itself, which is precisely the inversion most procurement processes get wrong.
Trust gets earned in stages. On a current engagement, AI is running in observation-only mode, locally, with no data leaving the network and access controlled by role, so the team can learn what it does and how accurate it is before it touches their work. Scope expands as results prove out, not on a vendor’s timeline. Holden also vets vendors against real constraints before engaging, ruling out one platform early because it was cloud-only while the client required everything on-premises, a decision he says saved months. The subtler risk is skill erosion. “Networks fail at 2 a.m. during storms, and someone has to understand the system well enough to fix it.” That is the case for keeping people trained, processes documented, and humans in the loop on what matters most. It is also the case for building depth instead of depending on a few people who carry critical knowledge in their heads, an exposure Holden sees across smaller and rural operators, where it works until someone retires. The leaders who thrive over the next three to five years, in his view, will distribute decisions closer to the network and the customer, demand line-level transparency from partners, and keep learning when programs hit trouble. The ones who fall behind will chase the next tool and measure results only after it is too late to change them.
Follow Timothy A. Holden on LinkedIn for more insights on program execution, operational accountability, and disciplined AI adoption in telecom.
