The moment I realized our onboarding model was broken wasn't a dramatic one. It was a quiet one.
A new customer service representative (CSR), straight out of five weeks of intensive training, was sitting at her desk when her first call came in. She knew the process; she'd been tested on it and passed with flying colors. She knew the policy. She'd read the scripts. And she froze. Not because the training hadn't been thorough, but because nothing in those previous five weeks had prepared her for the reality of being greeted by a frustrated customer with a complex query and the pressure of knowing that somewhere, someone was measuring how long she took to answer it. She fumbled, felt flustered, and in a wave of panic raised her hand for help.
That was my first experience of taking a live customer call in a contact center, 29 years ago. And I’ve thought about that moment a lot since.
What I realize now is that my lack of confidence on that first call wasn’t my failure. It was the system’s. Since then, I’ve spent over two decades leading and influencing customer service operations at scale across large enterprise organizations. And the thing that still strikes me is that, in all that time, the onboarding experience for CSRs has not fundamentally changed. That is a strategic risk.
The cost of getting it wrong shows up everywhere: reduced customer satisfaction, repeat contacts, higher abandonment rates, and high agent attrition. Contact center staff turnover remains one of the industry’s most persistent challenges, consistently benchmarked between 20–50%+ annually. But the real issue isn’t just the churn. It’s the onboarding operating model that creates it. What’s changed now is that AI gives us the ability to close this gap in the moment it matters.
What traditional onboarding gets wrong
The classroom prepares people for a world that doesn’t exist. As AI takes on predictable and repeatable work, the interactions that remain are more complex, more emotional, and less predictable. The bar for every human interaction has risen and continues to do so.
Despite this, most onboarding models haven’t adapted. Organizations invest heavily in structured training. Good training, delivered by capable teams. But when frontline teams hit the live environment, they still struggle. Why? Because the problem isn’t the quality of training. It’s the gap between training and real-world human engagement. No classroom can replicate:
- A frustrated customer who won’t follow the script
- A query that doesn’t fit the process
- The pressure of time, metrics, and expectation
CSRs aren’t failing because they don’t know enough. They’re failing because they’re expected to apply everything they have learned instantly and under pressure. And that’s where the model breaks.
The workarounds that never quite work
Historically, we’ve tried to solve this by giving CSRs more:
- Training
- Documentation
- Systems
- Support from experienced colleagues
But more information doesn’t equal better performance under pressure. In fact, the opposite often happens. We’ve trained teams to memorize where to click and which tab to search when what we should have been doing is focusing on building their confidence to think, respond, and connect in an empathetic way with the customer.
Instead, we’ve created environments where CSRs are juggling:
- Systems
- Processes
- Policies
- Customer emotion
…all at once. And when the pressure hits, they freeze. That’s the core issue. And it’s why incremental improvements to training have never fully worked.
Where organizations go wrong with AI
When AI first emerged in contact centers, most organizations layered it onto existing models:
- Knowledge bases
- AI-powered search
- Chatbots for contact center teams
All useful. None transformative. Because if your onboarding model produces CSRs who struggle to apply knowledge under pressure, giving them faster access to knowledge doesn’t solve the problem. It just means they struggle faster.
The same mistake appears when brands try to reduce ramp time by compressing training. Teams take calls earlier but confidence arrives much later. Speed without readiness isn’t efficiency. It’s a different kind of cost paid by the customer.
And this challenge doesn’t stop at onboarding. It repeats every time there is a training requirement:
- New product launches
- Policy changes
- Campaign goes live
- Team workflow changes
Each time, we rely on briefings and documentation and expect great performance to naturally follow. But performance doesn’t come from information alone. It comes from experience and confidence.
What actually works
Supporting performance in the moment.The real shift isn’t better training. It’s supporting performance while it’s happening. With Pega Customer Service™, AI doesn’t ask CSRs to search, remember, or navigate under pressure. It guides them in real time through:
- Next-best-action decisioning, adapting to the interaction as it unfolds
- Contextual knowledge surfacing, presenting the right information at the right moment
- Embedded workflow guidance, ensuring consistency without interrupting the conversation
These capabilities are powered by real-time decisioning, continuously recalculating what the CSR should do next based on the live context of the interaction. This fundamentally changes the role of the customer service representative.
Instead of managing systems and recalling policies, they can focus on what only humans can do well: being present, listening, understanding, and responding with empathy. But supporting performance in the moment is only part of the solution. The other part is preparing team members for the moment before it happens.
The Pega Customer Service Simulator
For decades, the closest we’ve had to real-world preparation has been role-play.
- A colleague pretending to be a customer.
- A scenario everyone knows isn’t real.
- No real pressure. No real consequence.
It’s better than nothing. But it’s never been enough. What organizations need is not better role-play. They need experience before live exposure.
The Pega Customer Service Simulator addresses that gap by creating AI-driven practice environments that reflect the variability and unpredictability of real customer interactions. Within this environment, CSRs can:
- Handle complex and emotional conversations before they face them live
- Experience scenarios that don’t follow a script
- Make mistakes safely, without customer impact
- Build instinctive responses through repetition
This is where confidence actually comes from. And importantly, this isn’t just for onboarding. Every time something changes – products, policies, processes – the Simulator provides a way to build readiness before go-live.
Bringing it together
What’s now possible is something the industry hasn’t had before:
A combined model where CSRs:
- Practice in realistic scenarios (Simulator)
- Perform with real-time guidance (AI decisioning + workflows
- Improve through experience, not just instruction
This shifts onboarding from a training exercise to a readiness strategy. And when that happens:
- Ramp time becomes more predictable
- Quality becomes more consistent
- Confidence builds earlier
- Attrition reduces because teams feel capable, not overwhelmed
Where to go from here
If any of this resonates, it’s worth asking a simple question: How long does it really take for a CSR in your operation to feel confident, not just trained?
And more importantly: What happens to customers in that gap?
I’d be interested to hear how others are approaching this:
- Where do you see the biggest gap today, training or performance under pressure?
- How are you building confidence, not just competence, in your teams?
If you haven’t seen what’s possible with simulation and real-time guidance working together, that’s a useful place to start. The difference isn’t incremental it’s structural.