You’d be surprised at just how many CEOs are picking up the phone and calling their own contact centers.
Not to check on staffing levels or review a transcript; just to experience it as a customer would.
There is a pleasure in discovering that, despite being surrounded by and well-informed about all the data and latest technological advancements in the contact center space, many CEOs are still resorting to the good old-fashioned methods.
Unfortunately, according to Claudio Rodrigues, Chief Product Officer at Omilia, the answer is rarely flattering:
“You will see many CEOs calling their contact centers just to see how the experience goes. And so many times they are disappointed.”
That disappointment points to something that the contact center industry has been slow to fully reckon with.
For years, the dominant frame for evaluating contact center performance has been cost, be it cost per call, cost to serve, or headcount reduction. AI arrived and largely inherited that framing. The result is a generation of deployments optimized to deflect volume… and not a lot else.
The leaders pulling ahead right now are working from a different premise. Instead of treating it as a cost line, they are treating it as the most direct, high-stakes interface a business has with its customers.
Treat it like a drain on the budget, and that’s exactly what it becomes.
What a Fragmented Stack Actually Costs You
The reason most contact centers haven’t made the leap from automation to genuine intelligence is a technology architecture problem.
A typical enterprise contact center runs on a collection of disconnected tools: a CRM holding customer data, a separate platform storing call recordings for compliance, a workforce optimization tool monitoring agent adherence to SOPs, and a knowledge base sitting somewhere else entirely.
Each one generates data, but very little of that data talks to the rest.
“They never get the full picture,” says Rodrigues.
“Each system understands one part of the customer. Nobody understands THE customer.”
The intelligence loop that would allow a contact center to learn from every interaction, spot gaps, and improve continuously requires two things to be true simultaneously:
- You need to be listening to both the agent and the customer at the same time
- You need a single place where that information lands and gets structured
Without that, the loop never closes. You may get local optimizations – a slight improvement here, a tweak there – but never the compound gains that come from genuinely connected data.
Rodrigues gives a concrete example of what broken looks like in practice, detailing an example where a customer is calling to upgrade their hotel room, and the agent is unable to action it in the system, resulting in a potential revenue moment slipping away:
“You can immediately understand just by listening to this conversation that you are putting revenue in jeopardy. And you could have been capturing it right at that moment.”
The Compounding Case for Self-Learning
When Rodrigues talks about ROI, he frames it less as a headline number and more as a compounding process – one that most platforms haven’t been built to support.
The opportunity, in his view, lies in the hundreds of small, recurring interaction types that were always technically automatable but never made the priority list because the manual effort required to build and maintain them wasn’t worth it.
With a self-learning system, those cases become easy wins that accumulate over time.
“If you can serve 15, 20, or 30% more calls just because of the compounding of small percentage points, you can finally automate because it’s effortless,” he says, “then your ROI has gone up, and you got it so much faster with so little effort.”
The practical lesson here for CX leaders is changing how they think about their AI backlog.
Things that have been sitting at the bottom of the roadmap for two years – not because they’re unimportant, but because they’ve never been worth the effort to build – can become viable in a system that handles the heavy lifting automatically.
The Mindset Shift That Precedes Everything Else
For leaders looking to make that move, Rodrigues is direct about where to start: know what you’re trying to achieve before you touch the technology.
“Businesses that define goals very well will achieve those goals, or know that they did not achieve them and change their strategy,” he says.
“That didn’t change before AI, and it is not going to change after.”
The more consequential question, once that clarity exists, is knowing where humans belong in the model.
Rodrigues isn’t making a case for replacing contact center agents. He’s arguing for what he calls human augmentation, where contact centers keep the best brand ambassadors on the front line and let AI extend their reach and capabilities, rather than substituting for them.
“Let AI augment the capabilities of those brand ambassadors hundreds of times over,” he says. “There is a convergence that does not leave humans outside of it.”
The contact centers that figure that balance out – clear goals, connected data, AI handling the long tail while humans handle what humans do best – are the ones that will make the revenue engine argument to their boards in numbers that actually stack up.
