Home > Columns > CRM Columns

The Four Jobs Every Agent Is Doing — and Why Your Organization Can't See All of Them

Contributed article by Mitch Lieberman

Every contact center agent is doing at least four jobs at once. As leads we do not give them enough credit for how hard their job is, and can be. They are managing a conversation, following a process, solving a problem, and operating a technology stack. Simultaneously, under time pressure, with a real person on the other end waiting for help. 

The skills vocabulary most organizations use to hire, train, and evaluate those agents accounts for only one of those jobs at a time. Sometimes two. Rarely all four, and almost never in a way that connects Quality Assurance (QA), training, coaching, and workforce management (WFM) to the same underlying picture of agent capability. 

That is the coordination failure at the center of contact center performance management. Not the tools, not the agents or the maps. This is a team sport. 

Different Functions, Different Lenses

QA scorecards measure communication and compliance, mostly soft skills. Training curricula cover product knowledge and call flows, while hiring profiles screen for empathy and multitasking. Workforce management routes (WFM) on tenure and queue type. Each function is doing its job. They are using different vocabularies to describe the same agent. 

When a QA trend shows "communication scores are dropping," that information doesn't automatically tell training which specific behavior to develop. It doesn't tell WFM which queue assignments to adjust and it doesn't tell hiring which screening criteria to tighten. The data exists, but the diagnosis doesn't.

SQM Group’s research on first call resolution (FCR) links directly to operating efficiency: for every 1% improvement in FCR, a call center reduces operating costs by 1%, and for the average midsize call center, that improvement is worth about $286,000 in annual operational savings. SQM also positions FCR as both a customer service effectiveness metric and an operating efficiency metric, because higher FCR is associated with fewer repeat contacts and lower service costs. 

For a center handling 40,000 interactions per month, that is roughly $300K–$500K annually per percentage point. Repeat contacts, mis-routed calls, and unresolved issues all trace back to skills gaps that organizations can measure but cannot diagnose — because QA, training, and WFM are not working from the same framework. 

What This Looks Like from the Customer's Side

Consider a customer who calls about a billing discrepancy. The agent is polite, uses the customer's name, and acknowledges the frustration. All solid communication behaviors. But the agent can't diagnose why the charge appeared. They search the knowledge base, find a generic article, and issue a credit without understanding the root cause. Three weeks later, the same charge appears. The customer calls back. A different agent picks up, sees no documentation of the root cause, and starts from scratch. 

From the customer's perspective, this is one experience: the company charged them twice and couldn't figure out why. From the organization's perspective, these are two separate tickets, both of which received passing QA scores. The first agent scored well on communication. The second followed the correct process. Neither was coached on what actually went wrong, because the skills vocabulary didn't connect the diagnosis failure to the documentation failure to the repeat contact. 

Qualtrics research consistently shows that high-effort customer experiences are the strongest predictor of churn. This is what high-effort looks like from the inside. It is not dramatic. It is corrosive. And it compounds. 

Four Domains, One Map

The solution isn't a new theory imposed on top of existing operations. It's alignment to structures that are already there. 

Academic and occupational research, drawn from O*NET classifications, Lightcast workforce analytics, and industry training frameworks, converges on roughly six distinct skill clusters in contact center environments. Six is analytically useful but operationally cumbersome. Contact centers don't evaluate agents across six scorecard sections. They evaluate them across three or four. 

The four-domain model consolidates those clusters into categories that match how QA platforms already work: Communication & Interpersonal Skills, Process & Compliance Execution, Knowledge & Problem-Solving, and Digital Dexterity & Systems Proficiency. Balto, Calabrio, Observe.AI, and Call Center Studio all cluster their metrics into groupings that map directly onto these four domains. Most organizations adopting this model can relabel or lightly restructure existing scorecards rather than rebuilding from scratch. The operational vocabulary is already there. The shared language is what's missing. 

Each domain names something distinct that can go wrong independently. An agent can be excellent at communication and still generate repeat contacts because their knowledge of root cause analysis is shallow;  they treat symptoms, not problems. An agent can know the product cold and still create compliance risk because they rush through disclosure steps on difficult calls. An agent can handle both beautifully and still burn 25% of every interaction on after-call work because their CRM navigation is slow. 

These are not the same failures. They require different coaching, different routing decisions, and different hiring screens. The four-domain model doesn't just name the gap. It tells you which gap you're looking at. 

The Payoff: Coaching That Names the Problem

When HR, Training, QA, and Operations all use the same four categories, data from one function is immediately readable by the others. 

An agent struggling with repeat contacts gets a coaching note that says "your Knowledge & Problem-Solving scores suggest you're treating symptoms rather than diagnosing root causes, so, here's a scenario to practice" instead of "you need to improve your resolution quality." That is not a semantic distinction. The first gives the agent something to practice. The second gives them something to feel bad about. 

The shared taxonomy also creates the diagnostic foundation that makes AI-assisted coaching genuinely targeted rather than generic. When a QA review surfaces a pattern; an agent consistently skipping identity verification, or repeatedly rushing through closing steps on difficult calls, a practice scenario can appear in that agent's queue within hours. Not two weeks later, when the behavioral memory has faded. 

Ebbinghaus's research on the forgetting curve, first published in 1885 and confirmed by a century of learning science since, makes the point plainly: coached content that isn't practiced within days has limited staying power. The traditional QA-to-coaching cycle like a call recorded Tuesday, reviewed Thursday, coaching scheduled for two weeks from Monday, works against retention by design. A shared skills taxonomy with targeted, triggered practice scenarios works with it. 

AI coaching is not a prerequisite for using the four-domain model. Organizations benefit from the shared vocabulary and alignment long before any AI is deployed. But once the taxonomy is established, AI coaching becomes the mechanism that compresses weeks of lag into hours and makes skills development continuous rather than episodic. 

The Useful Question

The model is a starting point. The domains can be adapted to specific industries, regulatory environments, and organizational structures. What matters is the shared language and whether a QA finding in one function can be acted on immediately by another. 

Which brings the question worth sitting with: when QA identifies a skills gap in your contact center today, how many steps does it take before a specific coaching intervention happens and how many functions have to translate the finding into their own vocabulary before it gets there? 

If the answer involves more than two steps and more than one translation, you have a coordination failure. The four domains give you a way to reduce that friction and make a QA finding usable across the operation.

From Framework to Practice

Implementation does not need to start with a full transformation program. It can start with a simpler question: are QA, training, coaching, and workforce management all describing agent performance in the same language? In most centers, they are not.

The first move is to take the scorecard you already have and map it to the four domains, then use those same categories across coaching, training, calibration, and routing. The goal is not to build a better framework on paper, but to make a skills gap identified in one part of the operation immediately usable by the others.

From there, the path is practical: sharpen onboarding around communication and compliance, build deeper diagnostic skills as agents take on more complexity, and shorten the gap between what QA identifies and what agents actually get to practice. That is how contact centers move from measuring performance in fragments to improving it as a system.

Mitch Lieberman is VP of Fuel iX, within TELUS Digital, and a CX strategist and operator focused on contact center performance, AI integration, and the organizational alignment that makes both work.