Delivering Feedback That Changes Behaviour
The evaluation is worthless if the conversation produces no change. What the research suggests, and what practice usually does instead.
A scored call becomes value only through a conversation. Most of those conversations follow a format that the evidence suggests does not work.
What usually happens
The supervisor reads the score. The agent hears a number and starts defending it.
Several items are covered, most of them minor.
The agent agrees in order to end the conversation.
Nothing changes, and the same item appears next month.
What the evidence suggests
Research on feedback effectiveness is consistent on several points that contact centre practice routinely violates.
Feedback about the task changes behaviour; feedback about the person does not. "The explanation of the charges was hard to follow" works. "You need to work on your communication" does not.
Specific beats general, by a wide margin.
Prompt beats delayed. Feedback a month after a call is about an event the agent cannot remember.
Fewer points beat many. One or two changes are actionable; six is a list nobody retains.
Feedback that threatens self-image produces defence rather than learning, which is why a score attached to pay makes the coaching conversation useless.
Self-assessment first improves reception. Asking the agent to evaluate the call before you do changes the dynamic substantially, and it is nearly free.
A conversation that works
Lead with the call, not the score. Listen to the relevant segment together.
Ask them first. "What would you do differently?" Agents identify most of their own issues, and something they identified is something they own.
One or two things. Choose the item most likely to affect an outcome, not the most points.
Be specific about the behaviour and its effect. What happened, what the consequence was, what to do instead.
Agree a concrete change that can be observed in the next call.
Follow up on that specific thing. Feedback with no follow-up teaches that feedback does not matter.
Keep the score out of the conversation where possible, or handle it separately.
Frequency
Short and often beats long and rare. Fifteen minutes weekly outperforms an hour monthly.
Delay is the enemy. A call from three weeks ago is an abstraction.
Where analytics enables it, feedback close to the call is a genuine advance — the agent remembers it, and the specific segment is available immediately.
Positive feedback
Under-used and not for morale reasons.
Identifying what someone did well tells them what to repeat, which is information they otherwise lack.
It has to be specific. "Good call" is noise. "The way you confirmed the timeline at the end is exactly why this one did not come back" is instruction.
A ratio weighted toward the specific and positive produces better reception of the corrective, and the evidence on this is reasonably strong.
When feedback is not the answer
A large share of QA findings are not agent problems, and delivering them as feedback is both ineffective and corrosive.
Knowledge gaps are a knowledge base problem.
Long silences are usually a systems problem.
Repeated identical errors across many agents are a training or process problem.
Compliance failures on a confusing script are a script problem.
Ask before every coaching conversation: is this the agent, or is this the operation? If several agents fail the same item, it is the operation, and coaching each of them individually is the least efficient possible response.
A fifteen-minute structure
A repeatable shape for the coaching conversation that fits in the time supervisors actually have.
Two minutes: what went well, specifically, with the segment played if possible.
Three minutes: the agent's own assessment. "What would you do differently?" Wait. Do not fill the silence.
Three minutes: one observation, with the behaviour, the effect and the alternative.
Three minutes: agree the change, concretely and observably.
Two minutes: check for anything blocking them — a system, a policy, missing information. This is where process findings come from and it is always skipped when time runs short.
Two minutes: confirm and record what was agreed.
Then follow up on that specific thing next week. A structure with no follow-up is a ritual, and agents identify it as one immediately.