V-Spark Online Help

Coverage and Hit/Miss scores

Two different types of scores are given to an application’s upper-level categories: Coverage and Hit/Miss. Detailed examples are provided in Scoring Examples.

Coverage Scores

The Coverage Score is based on the average score of all lower-level categories contained within the upper-level category. For example, if an upper-level category contains four leaf-level categories and three of them contain at least one match, but the fourth contains no matches, then Coverage Score is equal to the average of these scores, i.e. (1+1+1+0) / 4, or 75%. This Coverage score is stored in the V‑Spark database and is used to calculate Hit/Miss scores.

Coverage Scores are useful for analyses where every lower-level category is significant and qualifies to be matched, such as agent compliance evaluation. For example, if you have a category for Gathering Customer Data that contains lower-level categories for Account Number, Phone Number, and Birthdate and your agents need to collect all of this information, then it is important to determine how many of these three categories are satisfied. Examples for how this scoring works for individual calls versus daily or monthly collections of calls are available in Scoring Examples.

Hit/Miss scores

The Hit/Miss score is based on whether any matches are found in a category’s lower-level categories and is calculated using the Coverage Score. If at least one phrase matches for one lower-level category, then the Coverage Score is greater than 0%, and the Hit/Miss score is consequently equal to 100%. If none of the lower-level categories match, then the Coverage Score will equal 0%, and the Hit/Miss score will consequently equal 0% also. This dependence of scores is very similar to the binary Match/No Match values that are assigned to leaf-level categories during the calculation of Coverage Scores. Note that Hit/Miss scores are not stored in the V‑Spark database.

Hit/Miss scores are useful for analyses where a call need only match one leaf-level category, such as call driver analysis. If you have a category for Billing calls that contains one leaf-level category for Payments and one for Questions, you do not need to know how many leaf-level categories are matched by the call. You simply need to know that the call did match, because customers do not always call with multiple issues.