Most contact center leaders will tell you, if you ask directly, that they know exactly what's broken.
They know their call coverage is under 10%. They know QA data isn't making its way into coaching conversations. They know training completion rates are not the same thing as training impact. They just don't expect a solution to exist.
That's not complacency. That's resignation.
It's also one of the most consistent findings in new research we commissioned on the state of contact center performance management in 2026. We surveyed 109 Directors, VPs, and Executives running contact center operations across healthcare, financial services, insurance, software, and hospitality — through GLG, with no current ReflexAI customers included. We wanted to understand the state of training effectiveness, QA maturity, and AI adoption from the outside in.
What came back wasn't a picture of a market making steady progress. It was a market that has learned to live with a structural problem it doesn't believe is solvable.
This is what broken looks like
85% of contact centers have deployed at least one AI-powered tool for training or QA. That's not a market in early adoption. That's a market that has already bought in. And yet only 29% say they're effectively using AI in their operations. For every three organizations that deployed a tool, only one believes it's working.
That's not a technology gap. That's a deployment and integration gap.
Meanwhile: 58% say their training tools aren't fully meeting their needs. 61% say the same about QA tools. And 77% don't have enough time for value-add work — the coaching, calibration, and development work that would actually close the performance gap they're trying to close.
The tools are in place. The data exists. The performance improvement hasn't followed.
Where contact center training best practices break down
When we asked respondents to rate their training programs, the pattern was consistent across verticals and organization sizes: strong on paper, weak in practice.
85% agree they have clear performance standards. 79% say their training content is directly connected to those standards. These aren't organizations flying blind on what good looks like.
But only 57% say training provides enough realistic practice to prepare agents for live calls. Only 53% can effectively measure training's impact on agent performance. Only 49% have full visibility into trainee readiness before an agent takes their first live call.
The gap between knowing what good performance looks like and being able to improve call center agent performance consistently is where most organizations are stuck. They can't measure whether training changed anything once agents got on the floor.
The #1 unmet need isn't better training content. It's proof that training changes anything.
Where call center quality management falls apart
QA tells a similar story at a different stage of the workflow.
Data collection is strong. 86% track agent-by-agent statistics. 85% regularly review calls. The infrastructure to collect performance data is largely in place.
Everything after collection leaks.
Only 58% can easily see trends across their team. Only 59% can quickly act on those trends. Only 52% can measure whether their actions had any effect.
And even those numbers overstate reality — because most organizations are working from a sample of fewer than 10% of total interactions. The trends QA leaders are acting on, the coaching decisions they're making, and the performance gaps they believe they understand are all drawn from a fraction of what's actually happening on the floor.
The finding that surprised us most
We never asked whether training and QA were connected.
The survey didn't include a question about integration between functions. It didn't prompt respondents to rate the handoff between their QA team and their training team. And yet, across the open-ended responses, the training-QA disconnect surfaced repeatedly as the most critical gap respondents identified.
Multiple leaders described it as a white-space opportunity. One executive put it simply:
Nothing I've seen ties QA and training together, finding a tool that could would be great.
When a need surfaces unprompted, it's a more reliable signal than agreement on a structured question.
Training leaders can't see whether what they build changes on-floor behavior. QA leaders can't convert performance data into targeted development. Both sides describe the same broken handoff — and in environments where teams are already stretched thin, the manual coordination required to bridge that gap isn't realistic.
Need one tool to be a linchpin between training and QA. Focus should be on agent development rather than analytics.
Who's hurting most
The aggregate picture is challenging. The segment-level data is sharper.
Healthcare reports the lowest training tool satisfaction of any vertical (3.00/5) — but the frustration isn't just about tools. Healthcare respondents consistently describe regulatory constraints, compliance risk, and the tension between standardized scoring and clinical context. For a healthcare contact center, a mishandled call doesn't result in a CSAT dip. It carries clinical, legal, and reputational consequences that require a fundamentally different standard of training and QA accuracy.
Insurance reports the lowest AI effectiveness of any vertical (2.63) — lower than healthcare — despite having tools actively deployed. The technology is present. The value isn't.
The 151–250 agent segment is the most underserved across every dimension: training satisfaction (2.87), QA satisfaction (2.87), and AI effectiveness (2.60). These organizations have outgrown manual processes but haven't reached the volumes that attract enterprise vendors. They're caught in the middle — and most vendors aren't building for them.
What leaders actually want
When asked how new tools could impact their 2026 priorities, the responses converged around four things: automation of manual, repetitive work; 100% interaction coverage; real-time feedback; and a closed loop between QA findings and targeted practice.
The pattern is consistent. Leaders aren't asking for better training software or more comprehensive QA coverage in isolation. They're asking for a system that connects them.
Tools that capture 100% of interactions, structure them into data and turn that data into actionable insights would be the single biggest help for 2026.
The full report goes deeper
Segment-level data for healthcare and insurance, the complete 151–250 agent analysis, the full training and QA Likert breakdowns, and the open-end responses that surfaced the training-QA disconnect unprompted. If you're responsible for contact center performance management — or evaluating technology to support it — it's worth the read.
