Popular automations include sending automatic replies to tickets of a particular category — usually the most simple customer requests. Those automations are based on Cleverly’s Triage capabilities that label each ticket with a certain category.
The Cleverly Category has a certain confidence associated with it — for some tickets, the prediction model is more accurate than for others.
When an automation is configured based on the category, you have the possibility to define a confidence threshold so that the automation only runs if the model’s category prediction is of high confidence. By default, we consider the high threshold to be 70%, but that can be adjusted.
This will slightly decrease the number of tickets that get an automated reply, but you will have higher confidence in the answer being the right one.
To define the prediction confidence level you need to open the automation in question, or create a new one, and select the desired categories:
- The prediction confidence will show as a parameter of the “Category” condition;
- From the dropdown list, you can select between two values:
- Normal (default);
- High (70%).
If the automation is running for more than one ticket category, all category classifications will have to meet a level of confidence higher than or equal to the selected level. Selecting a higher confidence level leads to fewer tickets meeting this condition, but ensures that those which do are more likely to belong to the respective categories.