Happy employees make for happy customers. It's one of those business truths that gets repeated often enough to feel obvious, but the way most enterprises equip their customer service reps tells a different story.
Sam Stern, who leads Service Design & Customer Experience at LinkedIn, joined the PolyAI podcast and made the case that the experience your customers have is almost entirely downstream from the experience your employees have. Most enterprises aren't investing where the problem actually lives. Here's what that costs them, and where the real opportunities are.
Starting every conversation from zero
In most contact centers, customer conversations start from scratch. The agent who's just come off a difficult call picks up the phone again and immediately has to piece together who they're talking to, why they're calling, and what's already happened, all while the customer is waiting. They’re on the back foot before the call has even begun, working from incomplete information.
At LinkedIn, reps without context were speaking with customers who had accurate, real-time data in front of them because the business had shared it directly with them. The reps had a lagging, less accurate version, which left them building workarounds mid-call to reach the same starting point as the person they were supposed to be helping.
The toggle tax
Sam named the most consistent drain on employee focus as ‘the toggle tax’, when reps have to hop between the CRM, knowledge base, and product data to find information. While they do that, the call technically continues, but the flow of conversation and the human connection slowly evaporates. The customer almost always feels the lag, and the rep knows they're not as present as they should be. PolyAI’s research found that 63.5% of consumers say agents are only half-listening. In many cases, it’s because they’re multitasking throughout the conversation.
The fix shouldn’t be complicated. If you bring the data to the employee in one place rather than making them chase it, they have all of the relevant customer context in one place and can actually focus on the conversation. Layer AI on top of that integrated data to suggest next best actions, and you've gone from reactive problem solving to proactive relationship building. AI is often more accurate than a rep working from memory, but customers trust humans to deliver it.
When reps aren't chasing information across screens, they can actually show up for the conversation. That matters more as the interactions reaching them become more complex.
As human moments get rarer, they carry more weight
As interactions increasingly move to self-serve, the remaining human moments become more important. For many customers, a single interaction with your support or success team might be their only human contact with your business all year. When you get it right, they remember it. Service recovery is particularly powerful here because when something goes wrong and a human steps in, understands the situation, and resolves it clearly, that stays with the customer far longer than any digital interaction would. While the number of human touchpoints is falling, the weight each one carries is going up.
But you don’t just equip your employees once and move on. Your system gains new potency with every call when you combine human expertise with AI’s powerful ability to capture and distill information.
What those conversations should be adding up to
Every customer conversation generates signals of recurring questions and early indicators of what customers actually need. That should feed decisions, surface next best actions, and make your teams smarter over time. It rarely does because the systems to capture and use it aren't there.
At LinkedIn, Sam's team built an insights agent trained on years of research that any product manager can query directly without commissioning a new project or waiting on Sam's team. The voice of the customer is available in real time to anyone who needs it.
You have to build the infrastructure that makes conversations count, whether that's integrating insights into the tools your product teams use, making research available in real time rather than buried in a folder, or designing feedback loops that actually close. When you do, the voice of the customer starts informing every decision automatically.
Your conversations are already telling you what to fix
Most of the answers CX leaders are looking for already exist inside their conversational data. The problem is getting to them quickly enough to act.
PolyAI's Smart Analyst lets you explore millions of conversations through a simple conversational interface. Instead of waiting on BI teams or static dashboards, you ask a question and get an instant breakdown across hundreds or thousands of calls.
Teams have also used Smart Analyst to surface containment issues, catch data quality problems across dozens of sites, and find the root causes behind repeat delivery delays. The infrastructure Sam describes at LinkedIn, where customer insight reaches the people who need it in real time, is exactly what Smart Analyst is built to do.
Where most CX investment misses
Most CX investment goes into the customer-facing layer, and the employee experience underneath it gets left behind. If you get the balance right, better customer outcomes follow. Agentic AI is making that easier by automating the low-value tasks that eat into rep time, like CRM logging, and making them more valuable in the process.
If you want to understand why your customer experience isn’t delivering, start by asking your employees what's getting in the way. Sam covers that in more detail in the full episode, including what agentic AI enables for employee-facing tools and why Service Design belongs in product. You can listen here .
