Call center coaching that changes what happens on the next call
Gary Hale, Co-Founder & CEO
Most call center coaching sessions follow the same shape. A supervisor pulls up the month's quality scores, walks through the numbers, points out a low-scoring contact, and asks the agent to do better. The agent agrees. Both people leave. Nothing measurable changes.
The session was not useless because the supervisor was bad at it. It was useless because reviewing a score is not coaching. Coaching is changing a specific behavior on a specific type of contact and then verifying that it changed. Here is how to build that.
Coach behavior, not the score
A quality score is a summary. It tells you that something went wrong across a set of contacts. It does not tell an agent what to do differently on Tuesday.
The move is to convert the quality score into an observable behavior. "Your quality is at 82" becomes "on billing disputes, you are confirming the account before you have understood what the customer is disputing, and it costs you the rest of the call." The first is a grade. The second is an instruction an agent can act on immediately.
This requires the supervisor to have listened to contacts rather than only read scores, which is the real constraint in most programs. Budget the listening time explicitly. A coaching session built on three contacts the supervisor actually heard is worth more than one built on a month of aggregate scoring.
One behavior per session
The most common way coaching fails is volume. A supervisor who has saved up a month of observations delivers six pieces of feedback in one sitting. The agent remembers none of them, or remembers the harshest one and the general feeling that they are doing badly.
Pick one behavior. Make it the one with the largest effect on outcomes for that agent's contact mix. Everything else waits for the next session.
This feels slow and is not. One behavior changed per session and verified is twelve real changes a year. Six behaviors mentioned per session and none verified is zero.
Cadence beats intensity
Coaching that happens monthly is not coaching. The gap between observation and conversation is long enough that the agent does not remember the contact, and long enough for the behavior to be reinforced a few hundred times before anyone mentions it.
A workable cadence has three layers. Same-day feedback on specific contacts, brief and immediate, especially during the first 90 days. A weekly one-on-one, short, one behavior, with a verification step on the previous week's behavior. And a monthly review that looks at trend rather than incidents, where the quality score belongs.
The weekly session is the one that does the work, and it is the one that gets canceled. Protect it in the schedule. This is also a workforce management question, because coaching time is planned shrinkage and has to be built into the schedule rather than squeezed into whatever gap appears.
Verify the change on real contacts
A coaching session without a verification step is a conversation. The step is simple: after the session, pull contacts of the same type from the following week and check whether the behavior changed.
This does three things. It tells you whether the coaching worked, which most programs never actually learn. It tells the agent the standard was real. And it separates two situations that look identical on a scorecard: an agent who understood and changed, and an agent who agreed politely and did not, usually because they did not know how.
The second case is the important one. When verification shows no change, the next session is not a firmer version of the same message. It is a different diagnosis, usually a knowledge gap or an unclear process, and sometimes a tool that makes the desired behavior slow.
Calibrate, or coaching contradicts itself
If two supervisors score the same contact differently, agents receive contradictory coaching and learn that the standard is personal rather than real. Calibration is the fix and it is unglamorous: supervisors score the same contacts independently, compare, and resolve the differences until they converge.
Run it monthly and treat unresolved disagreement as a scorecard defect rather than a supervisor defect. If experienced people cannot agree on what a scorecard item means, the item is written badly and agents have no chance of hitting it consistently.
ISO 18295-1:2017, the international standard for customer contact centers, sets out requirements covering how centers manage and develop staff alongside customer-facing performance (International Organization for Standardization). Consistency of assessment is the foundation the rest of it rests on.
Coaching is a retention lever
Agents leave contact centers for reasons that coaching cannot fix, including pay and scheduling. It can fix one of the common ones: the sense that the work is unwinnable and nobody can explain how to succeed at it. That sits alongside the other drivers covered in our guidance on lowering agent turnover.
An agent who receives specific, weekly, actionable feedback and can see their own improvement is in a different job than an agent who receives a monthly score and a general request to improve. The Bureau of Labor Statistics projects roughly 289,500 openings per year for customer service representatives over the 2025 to 2035 decade, nearly all from replacement need (U.S. Bureau of Labor Statistics). Every avoided departure is a full training cycle you do not run again.
Mpathic hires on character, aptitude, and experience and runs operational discipline through key performance indicators, coaching, and quality assurance as a connected system rather than as separate functions.
Call center coaching works when it names one behavior, happens close to the contact, recurs weekly, and is verified on real contacts afterward. It fails when it reviews a score, covers six topics, happens monthly, and ends without a check.
Change one thing this month: add a verification step. Before each weekly session, pull two contacts of the type you coached on last week and check whether the behavior changed. You will learn quickly which of your coaching is landing, and that is the information the whole system has been missing.
Frequently asked questions
What is call center coaching?+
Call center coaching is the practice of observing an agent's contacts, identifying a specific behavior to change, discussing it directly with the agent, and verifying afterward whether the behavior changed. It differs from performance review in that it targets one observable action on a defined contact type rather than summarizing a period of scores.
How often should call center agents be coached?+
A weekly one-on-one focused on a single behavior, supported by same-day feedback on specific contacts during the first 90 days, and a monthly session for trend review. Monthly-only coaching leaves too long a gap between the contact and the conversation for the agent to recall the contact or for the behavior to change.
What should a coaching session cover?+
One behavior, on a specific contact type, described in terms an agent can act on immediately. Start by verifying whether last session's behavior changed, then introduce the new one. Leave quality score trends to the monthly review, since a score summarizes what happened rather than telling an agent what to do next.
What is calibration in call center quality assurance?+
Calibration is the process of having multiple supervisors or quality analysts score the same contacts independently, then compare and resolve their differences. It ensures agents receive consistent assessment regardless of who reviews their work. Persistent disagreement usually indicates a poorly written scorecard item rather than a supervisor problem.
Does coaching reduce call center attrition?+
It addresses one common driver, which is the sense that expectations are unclear and improvement is not possible. Specific, frequent, actionable feedback that an agent can see working changes that experience. It does not address pay or scheduling, which are separate retention levers and often the larger ones.
