Contact Center Pipeline November 2026 | Page 36

But automated quality gives teams a way to catch critical errors much closer to the moment they occur, so they can step in before the problem becomes harder to fix.

DS: All three of these pressures are real.

Formal feedback is drying up. Retently ' s study of more than 25 million eCommerce survey invitations found email response rates fell from 4.09 % in Q1 2025 to 2.50 % in Q4, even as survey volume nearly doubled. Surveying harder produces less signal, but the opinions still exist.

Sampling was always a limited view. Enterprise centers processing millions of interactions monthly still review under 5 %, according to the Retently email response survey.

Moving to full coverage takes more than a bigger scorecard. The objective shifts from grading a sample to understanding every conversation.

AI agents should be held to the same outcome standard as human agents, but measured with a different instrument.

AI agents can be tested at a scale no human process allows, running thousands of simulated conversations before a real customer is involved, then continuously monitoring live interactions for accuracy and compliance.

Remember, if an agent serves your customers, its quality is your responsibility, human or not.

-- ERIC WILLIAMSON

EW: There is definitely pressure to evolve QA, QM, and customer feedback methods as the traditional approach is no longer enough for how contact centers operate today.

Traditional QA programs often evaluate only a very small percentage of interactions, which creates an incomplete and sometimes unfair view of agent performance.

As more interactions move across digital, voice, self-service, and AI-driven channels, organizations need evaluation programs that are more comprehensive, objective, and tied to business outcomes.

Resistance to formal feedback often comes down to how it is introduced:

• If agents feel like technology is being used to watch over their shoulders, adoption will be difficult.

• But when organizations include agents in the rollout and position QA as a way to make feedback more objective, fair, and useful for career development, agents are more likely to see the value.

36 CONTACT CENTER PIPELINE

Additionally, although AI agents should be evaluated with the same level of rigor as human agents, they should not necessarily be judged against the same criteria.

An AI agent may need to be measured on containment, escalation accuracy, brand integrity, and how well it handles questions it should not answer.

However, human agents, especially those who are handling complex escalations, should be measured on empathy, resolution, accountability, and whether the customer’ s issue was truly solved.

Scorecards are still valuable, but only when they reflect what quality actually means for the business, the customer, and the agent.

The issue is not scorecarding itself. The problem lies with poorly designed scorecards that measure activity without accurately assessing performance, miss key areas of the CX, or push agents toward scripted behaviors instead of effective, human conversations.

When scorecards are optimized around the right outcomes, they become more than just a QA tool. They become a way to deliver objective feedback, personalize coaching, and help agents understand how to improve in ways that directly support better CX.

WHAT ARE YOUR RECOMMENDATIONS TO CONTACT CENTERS TO HELP THEM PROVIDE HIGH QUALITY CX, INCLUDING AND ENABLED BY EX?

AC: Organizations should decide what outcome they ' re asking AI to provide, then measure that one thing. From our research, 61 % of consumers now say they ' d rather talk to a human, and that number went up this year.

This isn ' t a verdict on AI, but on automation that gets three-quarters of the way there. So put self-service where it can finish the job end to end and put AI behind the agent everywhere else.

EX isn ' t a separate consideration; it’ s part of how the AI strategy works. Agents are telling us the work is getting harder: 94 % expect AI to change their role within three years, and 61 % expect more complex, technical work. Contact centers can ' t ask people to take on harder judgment calls while leaving the busywork on their desk. The companies that are doing this well get knowledge to the agents in the moment and automate manual processes endto-end. They use the time they recover for coaching instead of handle time targets, then feed what they learn back upstream.

CONTINUED ON PAGE 39