You listen to 3% of your calls.
You can score 100%.

Getting started

A Quality Score and a Coverage Rate Are Not the Same Thing

A quality score is only as valid as the calls it measures.

In a program that runs on samples, the quality score can rise while the coverage rate stays flat. If the calls nobody listens to contain a compliance breach, repeated misinformation or a cancellation signal, those findings never reach a report.

This guide covers CX Quality through its product screens. It shows how criteria are defined, which questions the dashboard answers, how the agent breakdown is read and how the coaching output is produced. The screens and figures in Section 05 come from the POC of a real customer operation.

Section 01

Coverage Rate: How Much of Your Calls Is Measured?

A quality score represents only the calls that were listened to.

An agent spends an average of 5,000 minutes a month on calls. Across the industry, roughly 3% of those calls are listened to and scored.

The rate stays this low because of capacity, not choice. A quality analyst spends an average of 8 minutes listening to a single call and filling in the form, and works 160 hours a month. In an operation running 10,000 calls a month, those two figures mean 8 full-time analysts would be needed to listen to every call. For the remaining 97%, no quality data is produced at all.

Section 02

What Are You Missing in the Calls Nobody Listens To?

Seven signals that exist only inside the conversation.

1. 01Compliance and Data Security Breach

   A mandatory disclosure is skipped, personal data is shared against the rules.

2. 02Consistently Skipped Criteria

   An agent skips the same criterion every day and nobody notices.

3. 03Repeated Misinformation

   A rule learned incorrectly is passed on to hundreds of customers the same way.

4. 04Churn Signals

   The sentence “I want to cancel” is said before the cancellation happens.

5. 05Repeat Contact Reasons

   A customer calling a third time about the same issue tells you what the first call missed.

6. 06Product and Pricing Feedback

   The most candid comment about the product is said on the phone and never reaches the product team.

7. 07Sales Opportunities Left on the Table

   The customer states their interest plainly and the opportunity is never recorded.

Section 03

The Arithmetic of Coverage

Beyond a certain scale, hiring stops being an option.

The calculation assumes 8 minutes of evaluation per call. It is a theoretical reference and cannot be used as a staffing plan.

Beyond a certain scale, full coverage is no longer a budget decision. It is an arithmetic limit.

Section 04

You Define the Criteria, the System Does the Evaluation

You write the criteria and update them whenever you want; the yardstick stays the same on every call.

An operation never has just one quality form: separate forms are used for inbound calls, cross-sell and collections, and each form holds dozens of criteria. In CX Quality, a criterion is defined by the 6 fields shown on the screen below.

123456

1. 1QuestionThe name of the criterion.
2. 2DescriptionDefines the behaviour expected from the agent.
3. 3Evaluation InstructionSets out when the criterion is met and when it is not.
4. 4Positive Example 1Shows what the correct behaviour actually sounds like on a call.
5. 5Positive Example 2Contains a different phrasing of the same criterion.
6. 6KeywordsLists the patterns to look for in the transcript.

The moment a criterion is switched on, 100% of calls are evaluated against it instead of 3%. Who performs the evaluation and when does not affect the result.

A Sample Criterion Set

The set below consists of criteria running in a real customer operation.

Criterion names, definitions and score weights are specific to your operation; CX Quality does not impose a ready-made template. Criteria considered critical can also be flagged: when one of them is skipped, the call score drops straight to zero and risky calls move to the top of the list.

Section 05

What Happened in a Real Operation?

POC data: 1,438 calls, 13 agents.

The screen below belongs to a real customer operation. At the start of the POC the quality processes were stable: the team was capable and supervisors ran their follow-ups regularly. Even so, the criterion average had stayed flat at 61.36%.

123456

1. 1Total CallsShows the 1,438 calls in the period.
2. 2Calls AnalysedEvery call was analysed; this is not a sample.
3. 3Agents EvaluatedAll 13 agents were measured with the same criterion set.
4. 4CX Quality ScoreGives the period average.
5. 5Daily Score TrendShows days below and above the threshold in different colours.
6. 6Trend by CriterionOpens each criterion's own curve underneath the overall score.

Why the average had stalled became clear once every call was analysed: specific agents were consistently skipping specific criteria, and sample-based evaluation had never shown it. The finding was spotted on the dashboard within the first week and the team acted on it.

The comparison below shows the same dashboard on 16 April and on 12 May. The process, the team and the criterion set are identical on both dates.

The clearest example is the criterion “Were the standard questions asked to the candidate for the appointment?” On 16 April it scored 36.42%; more than half of the agents were closing the call without asking the standard questions. Four weeks later the same criterion had risen to 77.62%.

16 April 12 May

Whichever date you drag the slider towards, that date's criterion list comes forward.

A call you do not measure is an operation you cannot manage.CX Quality showed what to fix in 4 weeks.

Section 06

Which Agent Struggles Where?

Is the problem across the operation, or in a handful of agents?

Once you know which criterion needs work across the operation, the second question follows: which agents are struggling with it?

1234

1. 1Agent ListEach row is one agent; the names are anonymised.
2. 2CX Quality ScoreRanks agents by their overall score.
3. 3The Criterion They Struggle WithShows how a single criterion is distributed across agents.
4. 4The Other Criterion ColumnsLists every criterion on the form on the same screen.

A single column tells you whether the criterion is low across the operation or specific to a few agents. In the example above, on the call opening criterion some agents sit at 0% and others at 72%. That distinction decides whether the action is classroom training or individual coaching.

Section 07

Agent-Level Coaching and the Right Action

Alongside the score, it also establishes which action belongs to whom.

After analysing every call, CX Quality does not just report the criterion score; it also produces a coaching output for each agent.

12345

1. 1Summary48 calls analysed, CX Quality score 94.78%.
2. 2Overall AssessmentSummarises which criteria the agent is strong on.
3. 3Coaching GoalStates which behaviour to change and how.
4. 4Resolution RateThe resolution rate on this agent's calls is 75%.
5. 5SentimentCustomer sentiment at the end of the call is 75% positive.

The overall assessment summarises which criteria the agent is strong on. The coaching goal defines which behaviour to change and how. In this example two goals are given: a clearer start at the opening, and a summary of the resolution at the close.

The same page also carries the agent's analysed call count, quality score, resolution rate and customer sentiment distribution.

Reading the Signal

Not every finding is a coaching topic at agent level. The same signal calls for a different action depending on its source, so the root cause is separated out before the coaching output is attached to it.

Section 08

Are You Tracking the Quality of Your Operation?

How many of the 15 statements below can you make today?

• We know our coverage rate.
• Our criteria describe observable behaviour.
• We track critical errors separately from the weighted average.
• We reviewed the form within the last 6 months.
• We know the scoring variance between evaluators as a number.
• We hold regular coaching sessions for every agent.
• We record and follow up on coaching sessions.
• We have a defined process for agents to appeal a score.
• We track the improvement trend criterion by criterion.
• We can see which agent struggles with which criterion.
• We derive root cause analysis from quality findings.
• We have measured the relationship between quality score and CSAT.
• Written channels are inside our quality scope as well.
• We measure outsourced teams with the same criterion set as internal teams.
• We report quality findings outside the operation on a regular basis.

Next step

See It on Your Own Calls

This is not a case study, it is your operation.

Only your own calls can show which criterion sits at which level in your operation and which agent struggles where.

Call Center Studio CX Quality

• Every Single CallNot a sample, 100% coverage.
• Consistent ResultsThe same criterion, the same yardstick on every call.
• Your Own CriteriaNo ready-made template; you write the form.
• Criterion and Agent BreakdownOpens up what sits beneath the overall score.
• Coaching RecommendationIt also states what to do next.
• Critical Errors Tracked SeparatelyA call that autofails does not blend into the average.

Call Center Studio is a cloud contact center infrastructure that brings voice, digital channels and artificial intelligence together on a single platform. It works with more than 700 organisations and serves customers in more than 65 countries, while maintaining a churn rate below 2%, which is considered low for the SaaS industry.

Questions

Frequently Asked Questions

The 8 questions asked most often when the solution is evaluated.

1. Can We Define Our Own Quality Form?

   Yes. You decide the criteria, the descriptions, the example sentences and the score weights; you do not have to fit a ready-made template.

2. Can We Update the Form Later?

   Yes. Criteria are updated from the form editing screen; you can add a new criterion, change an existing one or remove it.

3. Can I Flag My Critical Criteria as Autofail?

   Yes. Autofail is a flag on a criterion that invalidates the whole call when it is skipped; instead of deducting points it takes the score straight to zero. Items such as failing to read the mandatory disclosure or skipping identity verification are defined this way. Calls that autofail are reported separately from the weighted average and move to the top of the risky calls list.

4. Can I Measure Agent Performance?

   Yes. For each agent you see the score by criterion, the resolution rate and the customer sentiment distribution; agents are ranked by their overall score.

5. Can I Compare My Own Scores with the AI Scores?

   Yes. On the dashboard the quality score is tracked separately for AI and manual evaluation, and the two results can be compared over the same period.

6. Does CX Quality Coach My Agents?

   Yes. An overall assessment and a coaching goal are produced for every agent, and which behaviour to change and how is written out with call examples.

7. Can We See the Reason Behind the Score?

   Yes. Each criterion result is shown together with the part of the call it is based on, so no score is left unexplained.

8. Can We Test It with Our Own Calls?

   Yes. The demo runs on your own recordings and your own criteria, so you see the results before you decide.