QA scorecards have long shaped how contact center agents are coached, evaluated, and developed. Yet, according to Taylor Wittman, Senior Solutions Consultant at UJET, many quality programs still rely on criteria that no longer reflect how customers behave or what agents need.
In a CX Today interview with Rob Wilkinson, Wittman argued that 100% conversation analytics should change more than the volume of interactions a business can review.
The bigger opportunity is to make quality assurance (QA) more current, consistent, and useful for supervisors and agents.
That matters because many contact centers are now applying AI to legacy QA processes. If those processes are outdated, automation may simply help teams move faster in the wrong direction.
UJET Warns Stale QA Scorecards Can Scale Bad Habits
Many contact centers built their QA scorecards years ago. Those forms may still check for compliance, greeting standards, verification steps, and process adherence. But customer expectations, products, channels, and agent responsibilities have changed.
Wittman warned that AI can make stale QA programs look more efficient without making them any more accurate.
“If a scorecard hasn’t been updated in a very long time, you’re just grading inaccurately faster.”
That is the danger inside many AI-powered QA programs. A tool may complete more scorecards, save QA teams time or give leaders broader coverage, but if the scorecard measures the wrong behaviors, the business may reinforce outdated service habits across more interactions.
For agents, that can mean coaching toward an old version of the customer experience. For customers, it can mean that the issues driving frustration stay hidden behind compliance-style evaluation.
Why Grading At Scale Is Not Coaching At Scale
Wittman drew an important line between grading and coaching. AI can help contact centers score more interactions. However, a score alone does not develop an agent, change behavior, or solve the reason customers keep calling.
That distinction is especially important as vendors promote automated scorecards as a quick win. Wittman summarized the gap:
“Grading at scale isn’t necessarily coaching at scale.”
That should shape how CX leaders evaluate QA automation. A scorecard can tell a supervisor what happened. Coaching must help the agent understand why it happened, what to do differently, and whether the issue is even within the agent’s control.
If an escalation is caused by a product update, a broken billing process, or unclear self-service instructions, coaching the agent harder will not solve the real problem.
Full conversation analysis should help supervisors separate agent development opportunities from operational issues the wider business needs to fix.
Small Samples Can Distort Agent Development
Traditional QA often relies on a small sample of interactions, which creates both practical and cultural problems.
If an agent handles hundreds or thousands of conversations in a month, a handful of reviewed calls may not represent their broader performance. One difficult customer interaction can carry too much weight.
Wittman suggested that agents often learn to perform for the scorecard rather than focus on broader skill development. That can make QA feel like a test rather than a growth tool.
Full conversation analytics can improve the picture. Instead of judging agents on a tiny sample, supervisors can see patterns across every interaction, including strengths, repeat coaching needs, and context around difficult calls.
That fuller view can make feedback more credible. It can also reduce defensiveness, because the conversation is no longer built around one bad call. It is built around a broader performance pattern.
Fairness Is A QA Issue, Not Just A Technology Issue
Wittman cited workforce data suggesting that around 70% of frontline customer service roles are held by women. She then connected that workforce reality to how QA affects confidence, development, and progression.