The traditional playbook for customer experience success is becoming obsolete. As AI transforms how businesses interact with customers, the initiatives and metrics that guided CX strategy for decades no longer reflect the reality of AI-augmented service delivery.
Let me be blunt: AHT is dead. Or at the very least, its days are numbered as a priority CX success metric.
And that’s just one example of how the industry is shifting its priorities away from effort toward issue resolution—and prevention—in the age of AI.
AI unlocks human problem-solving potential
We’re witnessing a shift from customer-led interactions being failure-driven— where the customer contacts the brand because there is a problem—to interactions that are brand-led and value-driven, where the brand engages the customer before there is a problem or before the customer hits friction.
This proactive approach creates what I call “free capacity” for human associates. It’s ‘free’ not because it comes without cost, but because human agents are liberated to engage in creative problem-solving and exercise more judgment and empathy, rather than spend time on repetitive questions. When AI successfully handles routine inquiries through self-service and proactive outreach, organizations unlock their most valuable resource: human expertise applied to complex, nuanced customer needs.
But this promising vision comes with a critical caveat. The AI needs to work. And if it doesn’t, the handoff between AI and human associates must be seamless. If the handoff is poor, the customer is not going to want to use that channel again. They’re going to say, ‘I tried the bot, it didn’t work, just get me to a human.’ And then you’ve lost the opportunity to use AI effectively in the future with that customer. And that’s when AHT enters the picture again.
Pay back your knowledge debt to afford a modern data estate
Before organizations can successfully deploy AI at scale, they must confront what I call “knowledge debt”—the accumulated cost of years or decades of neglect in managing unstructured information. This debt manifests in outdated help articles, contradictory documentation, and information scattered across systems that no longer communicate with each other. It’s the unseen drag on AI performance that can undermine even the most sophisticated implementation.
Knowledge is a great place to start, as much of the information provided in customer service interactions comes from here, and there are tools to make quick work of knowledge data quality cleanup. But cleaning up knowledge bases is only part of the infrastructure equation.
Once out of knowledge debt, organizations also need to build a “modern data estate” made up of cloud-native platforms that centralize company data in AI-ready, readily accessible, low-latency systems. AI cannot easily or quickly access data in high-latency legacy systems, or where data must come from multiple systems for the same request. A nimble tech stack foundation will power a strong AI engine.
AHT replacements for the AI age
If AHT is dead, what should take its place? Flip the script to focus on customer effort across the journey, not just employee efficiency metrics within one channel like chat or voice. With a mix of AI and humans supporting customers, success comes from understanding Customer Effort Score - how hard is it for customers to get their issues solved with automation.
There’s also opportunity to analyze First Call Resolution (FCR) through a new lens:
- Did the AI correctly identify the customer’s main problem and route them to the right specialist?
- When customers have multiple issues, were transfers to different experts appropriate?
These questions acknowledge the complexity of AI-augmented service delivery while maintaining focus on customer outcomes. It allows organizations to focus on the friction points throughout the customer interaction that influence FCR, not just FCR itself.
In addition, use automation-related metrics that move away from productivity and toward outcomes. For example, Autonomous Resolution Rate (ARR) measures the percentage of issues solved by AI without human intervention. And Resolution Accuracy (RA) gauges how accurate an automated resolution is, as judged by human experts.
The workforce implications are equally profound. Fewer entry-level Tier 1 roles and a much wider middle of Tier 2 and Tier 3 skilled problem solvers are needed. Customer workforces need to shift from looking like a pyramid to looking like a diamond. This structural transformation requires not just hiring differently, but investing in upskilling existing teams and rethinking career progression paths.
The “Day Two” question
Perhaps most critically, I urge organizations leverage AI to ask the “day two” question: How will AI applications be managed and continuously improved after implementation? Following on, what metrics will keep ongoing management a top priority?
AI isn’t a project with an end date. It’s a capability requiring ongoing refinement, testing, and evolution.
For senior technology and customer experience leaders navigating this transformation, success requires confronting infrastructure limitations, reimagining success metrics, restructuring workforces, and committing to ongoing evolution rather than a one-time implementation.
The stakes are high, but so is the potential. Organizations that successfully balance AI capabilities with human expertise can fundamentally transform the value they deliver to customers.
This article draws from Rizzo’s recent insights shared with HBR Analytic Services. Read the full report, “Keeping Customer Service Human-Centered in an AI World,” at Harvard Business Review.
