Contact center AI ROI shifts toward resolution quality and governed execution
SiliconANGLE Cheryl Knight ● Covered by 2 sources
Contact center AI is being judged less by volume and more by real problem-solving. Analysts say the winning metric is now resolution quality, with governance and data doing the heavy lifting.
Based on reporting by SiliconANGLE, Cheryl Knight — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
The next round of contact center AI bragging rights may not be about how many calls a bot can swallow. It’s about whether the customer leaves with the problem actually fixed. That was the message from Bob Laliberte of theCUBE Research and Zeus Kerravala of ZK Research after their discussion on SiliconANGLE’s “The AI ROI in Contact Center Summit.” They came away from conversations with Cisco, Talkdesk, Zoom and Five9 with a common theme: end-to-end resolution, not just automation for its own sake.
Kerravala put it bluntly: “Resolution and resolution quality is the new unit of value.” He argued that agentic systems should be judged on whether a customer’s needs were completed across the full journey, not just whether a chat was contained or a call was deflected. That’s a useful correction. Too many AI deployments still celebrate the easy numbers and ignore the messier ones: satisfaction, effort, employee productivity, cost and growth.
The analysts also kept circling back to data quality and governance. If the underlying systems are fragmented or the knowledge is stale, AI doesn’t fix the operation. It just speeds up the mistakes. Kerravala’s warning was simple: a broken process can get to a bad destination faster. So the advice was to start with one clearly bounded, high-value workflow, set a baseline, and prove improvement before widening the scope.
Governance, in their view, can’t be a one-time gate before launch. It has to become part of the operating rhythm: constant testing, observability, policy enforcement and evaluation. And that is where the practical difference lies. If governance is done right, Kerravala said, it should help AI move faster, not slow it down.
The workforce piece matters too. The picture they sketched is not robots replacing the whole contact center, but people moving toward exceptions, emotional calls and judgment-heavy cases while AI handles the routine work. Laliberte’s bottom line was refreshingly unsentimental: ROI won’t come from the number of bots deployed. It will come from better resolutions, more capable employees and efficient operations done at scale, without the usual chaos.
My take — AI-written commentary, not fact-checked reporting
This is the right correction, and it should embarrass a lot of AI vendors. Counting bots is the tech equivalent of grading a restaurant by how fast it clears plates. The real test is whether the customer got fed, and whether anyone had to pretend the burnt edges were a feature.
Read more about this at: SiliconANGLE