What QA Does to the People Being Measured
Contact centre attrition is high and quality programmes contribute to it. The design choices that make monitoring tolerable are known and rarely applied.
A quality programme is experienced by agents as being watched and judged. Whether that experience is fair, explicable and useful determines a great deal about whether people stay.
What agents object to, specifically
Small samples used to characterise them. Four calls out of four hundred, and the resulting number affects their standing. The objection is statistically correct.
Evaluator inconsistency. Being scored differently by different people for the same behaviour, which is measurable and usually unmeasured.
Findings they cannot examine. A score with no access to the call.
Automated determinations with no explanation, particularly where the agent believes the system misheard them.
Being coached on things outside their control — a broken system, a bad script, a policy the customer dislikes.
Moving targets. A scorecard that changes without notice.
Feedback long after the event.
None of these is unreasonable and all of them are fixable.
What makes monitoring tolerable
Transparency about what is monitored, including the limits.
A dispute route that sometimes upholds disputes. This is the single strongest signal that the programme wants to be accurate rather than unchallenged.
Access to their own data. The calls, the scores, the reasoning.
Findings routed to the right owner. An agent who sees a systems problem fixed because QA found it has a different relationship with the programme than one who is coached on it repeatedly.
Coaching separated from scoring, so the development conversation is not a negotiation.
Consistency, which requires calibration.
Recognition of good work, specifically, not as a morale exercise but as information about what to repeat.
The attrition connection
Contact centre attrition is high for many reasons and quality programmes are among them.
Agents leave managers and measurement systems, and a badly designed QA programme is both.
The cost of attrition is large — recruitment, training, the productivity ramp, and the quality cost of an inexperienced floor.
Any QA improvement that reduces attrition pays for itself at a rate that dwarfs the value of the scores.
This is worth putting in the business case for calibration, dispute handling and process routing, because those are otherwise seen as overhead.
Measuring the agent experience of QA
Rarely done and straightforward.
Ask, anonymously: do you understand how you are evaluated, do you think it is fair, does the feedback help, do you know how to dispute.
Track the answers over time, and against attrition.
Track dispute rates, which fall to zero when agents give up.
Track the proportion of evaluations actually delivered, which is a measure of whether supervisors have time.
A programme scoring well on quality metrics and badly on these is producing numbers at the expense of the people producing the service.
The surveillance threshold
There is a point at which monitoring changes the nature of the work.
Continuous analysis of every word, with real-time alerts and emotion inference, is a different job from one where a sample is reviewed.
Agents describe the difference, and the research on monitoring and wellbeing supports it.
The question worth asking of any new capability: does this improve outcomes enough to justify what it does to the working environment?
For compliance presence checking, usually yes. For continuous emotional assessment, the answer should be no on the evidence, and the note on emotion detection explains why.
The design principle
Monitor the work, not the person.
Analytics applied to processes, queues and systems produces findings that help everyone. The same analytics aimed primarily at individuals produces a workforce that manages the measurement rather than the customer, and leaves.
The anonymous check
Six questions, asked anonymously, twice a year, that measure how the programme is experienced.
Do you understand how you are evaluated?
Do you think the evaluation is fair?
Does the feedback help you do the job better?
Do you know how to dispute a finding?
Do you believe a dispute would be considered honestly?
Is there anything you do differently because you know calls are analysed?
That last question is the most informative and the most uncomfortable. Answers describing defensive or scripted behaviour indicate the programme is changing the work in ways nobody intended.
Track the answers against attrition and against QA scores. A programme with rising scores and falling answers here is optimising the measurement at the expense of the operation.