Automated keyword tagging

Learn how Sonar tags your experience data with keywords to save you time and ensure consistency

Niclas Nielsen avatar
Written by Niclas Nielsen
Updated over a week ago

Keyword tags are an essential asset on Sonar as they enable you to filter and categorise all your experience data. A keyword can be a subject, theme or trait that adds contextual meaning to a quote or an observation and enables you to group your CX within a study or over time.

Automatic keyword tagging is a new feature on Sonar and it has been made possible through a proprietary AI model that has been trained on years of historical keyword data, manually created by Sonar's network of Human Insights Specialists. The model has also been trained on specific local languages to achieve a high similarity on semantically equal sentences across languages.

When you, as a Producer Pro user, analyze your CX data, the keyword tags will automatically be generated for you to review and easily adjust - and every time you add or remove a keyword, the model will learn and continuously improve its accuracy. You can view and edit keyword tags both from within participant interview transcriptions, or when working with your nuggets in the Quotes & Observations or Insights pages.

Automatic keyword tagging will not only save you and Sonar's Human Insights Specialists a lot of time, it will also secure a much higher consistency (=quality) in the data, making it easier for you to generate insights.

Manage keywords

You can see a list of all keywords contained in a study, under the Quotes & Observations tab. Simply, click on the Project Keywords button in the top right. Here you can also add and remove keywords.

If you have assigned a new keyword to at least 5 quotes or observations, you can click the Assign Keywords (Automatic) button, next to the Project Keywords button. This option will assign the new keywords to existing Quotes & Observations, where the context matches with the Quotes & Observations already containing the keywords.

Did this answer your question?