Reps were failing because the environment was designed wrong, not because they did not know enough. I changed the environment.
New hires sat through four weeks of product and pricing training across a complex portfolio, then hit the phones cold. Their first live calls included every possible product combination and customer request type. No controlled practice, no progressive complexity, no pattern recognition. Every call was a new puzzle. Ramp to 100% quota attainment averaged three months post-training, and the conventional explanation was that the portfolio was just hard to learn.
The portfolio was not the problem. The training architecture was. Reps were being given all the knowledge upfront and none of the environment to apply it. I restructured the model around a learn/apply cycle: one product at a time, with live phone practice built in immediately after each module. But the model only worked if reps could practice on the right calls. That required a configuration change in the ACD (automatic call distributor) to route only single-product customers to new hire queues. I coordinated with technology and operations to set up the specialized skilling and routing rules within existing infrastructure. No new build. No new system. Just a configuration that made controlled learning possible.
From there, I layered in progressive complexity (gradually introducing multi-product interactions), integrated supervisor coaching earlier in the cycle, and built a repeatable coaching framework that frontline managers could run independently. The model was validated across three locations, two cohorts per location, with consistent results across all geographies. Ramp to 100% quota attainment dropped from three months to two months. The model remained the standard after my departure because it was built to run without ongoing facilitation.
The ability to diagnose structural problems that look like content or capability gaps, design system-level solutions that change the operating environment, coordinate cross-functional implementation, and build models that sustain without the person who created them.