If the L&D Team Disappears, What Happens to Training?

The senior L&D function is changing faster than most organizations realize. Recent data reported by HR Executive found that 96% of companies surveyed had eliminated a senior learning and development position within a five-month window. Over that same period, training content created by frontline managers increased 4.4 times.

The training function is not vanishing. It is being redistributed. 

This pattern extends beyond L&D specifically. Broader workforce data shows management-only roles declining globally by more than 6% over the past three years, with organizations across industries flattening hierarchies and redistributing responsibilities downward. The logic is straightforward: reduce layers, push accountability closer to the work, use technology to handle the coordination.

What disappears with those roles is not immediately visible. When a centralized L&D team owns training design and delivery, there is a continuous force pushing toward a consistent baseline across locations, teams, and cohorts. The same scenario was trained the same way everywhere, because development of that training scenario occurred in an identifiable hub. When that ownership transfers to individual managers, training becomes a function of each manager's capacity, priorities, and interpretation of what matters. Consistency degrades not because anyone is necessarily doing it wrong, but because no one is responsible for coordination and consistency. 

Training documentation can follow the same pattern. In many organizations, L&D teams maintain records of what training happened, what it covered, and whether it connected to measurable outcomes. Not every organization did this well, but the function existed. When training becomes part of managerial responsibility, the impact of training becomes harder to prove in a metrics-driven way that resonates with senior leadership. The organization loses the ability to demonstrate that training occurred, that it was current, and that it considered the organization’s strategy alongside the operating reality on the front lines. In regulated industries, this diffusion can occur alongside an acute compliance risk. In all industries, leadership can lose visibility into the ROI of training. 

A loss that compounds most over time is the feedback loop between performance data and training design. In a high performing organization, quality assurance findings, customer feedback, and operational metrics inform what gets trained next. When QA identifies a pattern of agents struggling with a specific scenario, the L&D team builds a response. When that intermediary disappears, the feedback loop breaks. QA generates data. Managers receive it. But the translation from insight to intervention is inconsistent and ad hoc. In particular, realistic practice and simulations are likely to suffer.

Research we conducted earlier this year illustrates this directly. We found that 86% of organizations regularly review recorded interactions. But only 52% can measure whether their response to what they found actually changed performance. The measurement and accountability functions of senior L&D roles have not been automatically replaced by others. 

The instinct in response to these cuts is to point to AI as the obvious successor. AI-powered tools can generate training content, score interactions, deliver feedback, and automate quality evaluation at scale. 

The reality is more complicated. Within contact centers, our research found that 75% have deployed AI quality assurance tools but only 29% report that AI is effectively integrated into their operations. The distance between those two figures is telling. Deploying an AI tool that automates scoring or generates training scenarios addresses the production bottleneck. It does not address the systems design problem: connecting what performance data reveals to what training actually covers.

A manager using most AI tools to build a training module is still making decisions about what to train, when to train it, and how to evaluate whether it worked. Those decisions require the same contextual judgment that senior L&D roles provide. The tool accelerates the work. It does not replace the infrastructure that made the work coherent.

This is not an argument against AI in training. It is a diagnostic observation about what AI alone does not solve, particularly without the best tools. Organizations that deploy AI tools without connecting them to a feedback loop between evaluation and practice end up with faster, more scalable versions of the same fragmented approach. More things get produced, but less of it connects to anything.

What lean teams actually need is not more tools. They need training infrastructure that operates as a system rather than a collection of disconnected activities. Three characteristics distinguish the organizations that are closing the performance gap from those where training quality is quietly declining.

The first is training that updates from live data rather than quarterly curriculum reviews. When quality assurance identifies a pattern – agents struggling with a specific objection type, compliance language drifting from approved phrasing, a new product generating unfamiliar questions – the training system should be able to generate targeted practice scenarios from that insight without requiring a human to design, schedule, and deliver a response. The cycle time between identifying a problem and addressing it is the single largest determinant of whether training keeps pace with operational reality.

The second is practice that is auditable at scale. Every completed simulation, every assessment score, every progression metric is captured automatically, not because a compliance officer requires it, but because the organization needs evidence that training is changing behavior. The managers who inherited L&D responsibilities do not have time to manually document training activity. The infrastructure should do it for them.

The third is a closed loop between evaluation and practice. This is the element most organizations are missing and the one that senior L&D roles used to provide through institutional knowledge and manual coordination. In a closed-loop system, quality evaluation identifies specific performance issues, then uses those findings to inform the design of targeted practice scenarios. Agents complete those scenarios with measurable outcomes, and the system tracks whether the targeted practice actually improved performance on the original metric. Each step feeds the next without requiring a dedicated intermediary to translate between functions.

One financial services contact center we work with illustrates this pattern clearly. Their training lead uses automated quality evaluation to flag specific call behaviors: premature disqualification of qualified leads, weak objection handling, inconsistent compliance language. Those findings feed directly into AI-generated simulation scenarios that agents complete before taking live calls on those same topics. The training lead did not need a senior L&D team to build or maintain this loop. The infrastructure handles the translation between what evaluation finds and what training addresses. The result is a lean team operating with consistency and auditability.

The managers absorbing L&D responsibilities do not necessarily ask for this work and are not typically given additional resources to do it well. Expecting them to manually replicate what a dedicated team maintained is not a realistic strategy.

The answer is infrastructure that runs without a dedicated intermediary: training systems that update from live data without requiring someone to translate QA findings into curriculum, documentation that captures outcomes automatically rather than depending on managers to maintain records, and a feedback loop between evaluation and practice that operates as a system rather than a coordination task. Every team deserves access to the tools that will drive long-term, measurable success. 

The 96% of organizations that cut senior L&D positions made a resource decision. What happens next will determine the impact of that decision on the organization’s bottom line. The question is whether organizations will invest in the infrastructure to meet it before the training debt compounds into performance problems that are significantly more expensive to fix.

---

Sam Dorison is Cofounder and CEO of ReflexAI, where he builds AI-powered simulation training and quality assurance tools for high-stakes contact centers and behavioral health organizations. Before ReflexAI, Sam served as an executive at The Trevor Project, where his team's technology was recognized in TIME's 100 Best Inventions and MIT Technology Review. He writes about the intersection of AI, workforce development, and conversation performance.