Plan requirement
| Subscription | Any plan |
| Also required | Quality Assurance or Workforce Engagement Management |
| Access | Admin |
Keeping your rating categories useful. Two patterns tell you which to change: everyone scoring the same, and disputes clustering on one category.
Review them
- Go to Settings > Scorecards, then click Categories at the top.
- Look at the score distribution for each.
- Look at which generate disputes.
- Rewrite, retire or keep.
Pattern one: everybody always passes
A category where every review scores the same is measuring nothing. Either the behaviour is universal, in which case retire it, or the bar is too low to be informative.
It is also costing reviewer time on every review.
Pattern two: reviewers disagree
Disputes and calibration differences cluster on ambiguous categories. That is a wording problem, and rewriting the guidance fixes it.
Leaving it produces scores that depend on who reviewed, which agents notice quickly.
Deactivate rather than accumulate
Categories are added when something goes wrong and rarely removed. After two years a scorecard has eleven categories, reviews take fifteen minutes, and fewer of them happen. Retiring one is as legitimate as adding one.
Do it by deactivating. A deactivated category remains available in reporting and filtering, and your dashboards keep showing accurate historical scores and trends. Deleting is permanent and erases all category data from reporting, which takes a year of scores off every dashboard that used it.
Changing a category changes the meaning, not the data
Worth separating those two, because people avoid useful edits out of a fear of the wrong one. Editing a category's name and description does not affect its reporting, and changes apply to future conversations only. Nothing already scored is rewritten.
What does change is what the scores mean. If the rewritten question asks something materially different, the trend line spans two standards while looking continuous. Note the date on your reporting; that is the real risk, not data loss.
Ask reviewers what is hard to score
They know which category makes them hesitate. That is a faster route to the problem than any analysis of the distribution.
Once or twice a year
Enough to keep it accurate, infrequent enough that a year of data remains comparable.
Comments
0 comments
Article is closed for comments.