Get an immediate response from AI—no form required. Would you prefer to speak with an expert? Schedule an appointment.
Le Quality Monitoring consiste à évaluer, avec une grille de critères, un échantillon des interactions entre vos conseillers et vos clients, pour mesurer la qualité de service et la faire progresser. La définition tient en une phrase ; la difficulté est ailleurs : dans l’organisation. Cet article décrit les cinq étapes d’une démarche de Quality Monitoring qui tient dans la durée.
For a detailed definition and frequently asked questions, see What is Quality Monitoring?
1. Define the scope and objectives
Une démarche de Quality Monitoring échoue rarement sur la grille ; elle échoue sur l’ambition initiale. Avant d’écrire le moindre critère, fixez trois choses.
- The scope. Which activities, which channels, which sites? It’s better to start with one activity and one channel (phone support, email complaints) than to try to cover the entire customer service operation all at once.
- The objective. Regulatory compliance, first-contact resolution, customer service approach, adherence to procedures: each objective calls for different criteria. A system that tries to measure everything measures nothing.
- Resources. Who conducts the evaluations (supervisors, quality assurance team, service provider), how many hours per week, and how many evaluations per counselor per month. Four monthly evaluations per counselor is a common frequency; if the number falls below two, the results become meaningless.
Documentez ces choix dans une charte qualité d’une page, partagée avec les conseillers. Une démarche de Quality Monitoring n’est pas un contrôle surprise : plus il est explicite, mieux il est accepté.
2. Develop the rubric and calibrate the raters
The rubric translates your standards into observable criteria. A good criterion describes a behavior that can be observed during the interaction, not an impression. “The advisor restates the need before proposing a solution” is verifiable; “the advisor listens attentively” is not verifiable.
Organize the rubric into sections (introduction, comprehension, processing, conclusion, compliance), with a weighting for each criterion and three possible responses: compliant, non-compliant, not applicable. Reserve deal-breaker criteria (missing legal notice, unauthorized commitment) for an “unacceptable” category that reduces the score to zero, rather than diluting the issue by assigning a weight to it.
A rating scale is only valuable if two evaluators assign the same score to the same interaction. This is why calibration is necessary: before launch, and then every quarter, all evaluators rate the same three or four interactions, compare their results criterion by criterion, and adjust the wording of those that differ. The variation among evaluators is measured and must be tracked as an indicator in its own right.
Version the rubric. Each modification creates a new version, and evaluations made using two different versions should not be compared. For details on configuration options, see “Customizing Evaluation Methods” at Listener QM.
3. Organize Sampling and Quotas
We don’t evaluate everything; we evaluate what matters. The sampling rule determines which interactions are presented to the evaluators and in what proportions. It is defined by several variables: the team, the advisor, the reason for contact, the channel, the time period, and, if applicable, the duration.
- Quotas per counselor, not per team: otherwise, the most visible counselors are overrated and the others are overlooked.
- Explicit priorities: new customers for three months, sensitive issues (complaints, cancellations), and campaigns in the launch phase.
- An automatic assignment among evaluators, so that an advisor is not always evaluated by the same supervisor.
- Progress tracking: By mid-month, each evaluator knows how many evaluations they have left to complete.
Avoid “manual” selection in the telephony tool: it produces samples that are biased toward short, recent calls. A sampler applies the rule without exception and keeps track of what was selected and why.
4. Debrief and Turn Findings into Action
This is the step that is most often skipped—and the one that makes all the difference. An evaluation that isn’t followed up on in the days that follow loses most of its impact: the counselor no longer remembers the interaction, and the rating becomes just an abstract number.
The format that works: The advisor first self-evaluates using the same rubric, and then the debrief compares the two assessments. The discrepancy between the self-evaluation and the assessment is often more insightful than the score itself. The debrief concludes with a decision: an improvement plan with a time-bound goal, an assigned corrective action, or training focused on the criterion where the discrepancy was found.
The advisor must be able to challenge the assessment. A challengeable, well-documented assessment is accepted; an imposed assessment is simply endured. Build the possibility of challenges into the process, and view accepted challenges as a calibration signal.
For information on conducting debriefs, see “The Importance of Debriefing” at Quality Monitoring. For information on following up on actions taken, managing corrective actions, and scheduling training based on evaluations.
5. Mesurer et faire vivre la démarche de Quality Monitoring
A monthly quality review is sufficient, provided you focus on the right metrics and track them over time.
- The average score by criterion is more useful than the overall score: it’s what tells us where to provide training.
- The compliance rate by team, site, and channel, to compare like with like.
- L’écart inter-évaluateurs, qui mesure la fiabilité de la mesure elle-même.
- The time between the evaluation and the debrief, and the percentage of actions completed.
- The link to customer satisfaction: if the measured quality increases but satisfaction does not follow suit, the metric is measuring something other than what the customer actually feels.
Finally, review the evaluation grid once a year, and whenever there is a change in procedure or offerings. An evaluation grid that hasn’t been updated in three years measures customer service as it was three years ago.
When sampling becomes the limiting factor—that is, when discrepancies occur outside the scope of what raters can listen to—the process can be extended through automated “Quality Monitoring ,” which applies the same evaluation criteria to a much larger volume of conversations and leaves the cases that warrant human review to the raters.
Outiller la démarche de Quality Monitoring plutôt que la subir
Listener QM, CrossCX’s software Quality Monitoring, covers each of these five steps—versioned grids, calibration, sampling, debriefing and appeals, corrective actions, and quality reports—across all your channels.
