Call Labelling

In CallStudio, you can label calls manually for the purpose of improving your products and services.

What can I do with call labels?

There are many ways you can utilise call labels such as ML development or divide-and-conquer scenarios.

Using call labels for ML

Labelling calls is a manual process that takes time and effort. However, overtime the team can accumulate a volume of labels which can be used for developing and integrating new ML algorithms into the system. For instance, the team can give calls a new label needs investigation. In this scenario, ML algorithms can be used to automatically flag calls that should be investigated. This can help the supervisors identify issues at early stages before they develop.

Statistical analysis & comparison

The team can organise the calls into different populations by using labels. This can help data scientists analyse statistical differences in means or variables based on certain criteria (e.g., talking duration, talking speed).

Filter criteria construction

The team may wish to automatically classify calls based on simple filters or criteria. The labelling process can help identify decisive criteria that can easily flag certain labels. This requires a statistical analysis on the label populations and drawing decision boundaries. This criteria can be added into the filters in the Exploration Playground. This allows you to create dashboards that track different populations or subsets of your calls.

Automation criteria construction

Similar to the previous example, you may carry out the same process to build a set of criteria that can fire different automations such as alerting or flagging.

Labelling services

We provide crowd-based services where an external team can label a large volume of your calls quickly. This process is heavily controlled ensuring the best data quality possible.