This article helps you understand how to interpret and use the different Reviews dashboard cards in Zendesk QA to evaluate agent performance and identify areas for improvement.
This article contains the following topics:
- Understanding rating categories
- Calculating Scores by category
- Understanding individual review scores
- Calculating IQS
- Calculating category scores
- Calculating CSAT
Related articles:
Understanding rating categories
When setting up rating categories, your team will assign different weights to each category. Critical categories have a weight of 0.05 because they are pass/fail. See Deciding what rating categories to use.
| Critical categories | |||
| GDPR* | 0.05 | ||
| Language* | 0.05 | ||
| Soft Skills | |||
| Customer Facing | Politeness | 1 | |
| Explanation | 2 | Empathy | 0.5 |
| Solution | 2 | ||
| Business | |||
| Process | Upsell | 1 | |
| Ticket data | 1 | Self-help | 1 |
| Summary note | 1 |
Calculating scores by category
The highest possible rating for an agent is always 100% because the score is calculated based only on the rated categories, excluding any skipped categories.
The Scores by category card compares scores by category per agent. It doesn’t compare conversation averages for agents but category averages instead.
Each column displays the average category ratings given to or received by a specific agent. The average column calculates the overall value of these averages. Differences may occur when not all reviews contain the same number of ratings.
Understanding individual review scores
To find the exact conversation scores for an agent, use the Scores by reviews card.
If you want to find the category scores for your agents, independent of fail categories, use the Scores by category card. The main difference between the cards is that a negative fail category won’t result in a zero score for the entire conversation. This allows you to analyze categories across agents, independent of errors in fail categories.
Example of individual review scores
For example, consider the following scenario with five categories. Agent A received the following ratings:
| Review | Request* (0.05) | Clarification (2) | Explanation (2) | Writing (0.5) | Internal data (1) | Review Score |
| 1 | 100% | 100% | 100% | 100% | 100% | 100% |
| 2 | 100% | 0% | 0% | 100% | 0% | 9.91% |
| 3 | 100% | 0% | 100% | 100% | 100% | 63.96% |
| 4 | 100% | 100% | 100% | 0% | 100% | 90.99% |
| 5 | 0% | 100% | 100% | 100% | 100% | 0% |
The review score is calculated as follows:
review_score = (cat1_score * cat1_weight + cat2_score * cat2_weight + ...) / (cat1_weight + cat2_weight + ...)
In this case:
review_score = (request_Score * 0.05 + clarification_score * 2 + explanation_score * 2 + writing_score * 0.5 + internal-data_score * 1) / (0.05 + 2 + 2 + 0.5 + 1) >> unless request_Score < 50%, then 0%
In the Reviews dashboard, these numbers are rounded.
The review score for conversation five in the example above is zero because the fail category automatically sets this to zero.
Taking these same reviews together as category scores gives the following data.
Agent A:
| Request* | Clarification | Explanation | Writing | Internal data | Average | |
| Agent A | 80.00% | 60.00% | 80.00% | 80.00% | 80.00% | 76.00% |
The average score for Agent A is shown across all rated categories. It’s calculated as the average of the row, meaning the average scores per category for that agent were counted. The average per category highlights opportunities for improvement more explicitly than the average per conversation.
Calculating IQS
IQS, or Internal Quality Score, is derived from your conversation reviews. The average of all your reviews is calculated as follows:
IQS = (review1_score + review2_score + ….) / number of reviews
For example, consider the following review scores scenario:
| Review | Review Score |
|---|---|
| 1 | 100.00% |
| 2 | 9.91% |
| 3 | 63.96% |
| 4 | 90.99% |
| 5 | 0% |
| IQS | 52.97% |
IQS = (100% + 9.91% + 63.96% + 90.99% + 0%) / 5 = 52.97%
Calculating category scores
Refer to the table below to understand how category scores are calculated.
Calculating CSAT
Use the following guidelines to understand how CSAT (customer satisfaction) is calculated.
- Binary scale (or Good, Bad) 100, 0
- 3-point scale 100, 50, 0
- 4-point scale 100, 66, 33, 0
- 5-point scale 100, 75, 50, 25, 0
Divide the sum of all responses by the sum of total possible maximum scores.
