Skip to main content

Introducing the Sentiment engine

Charles Jorgensen
Updated

Introducing the Sentiment engine

Each individual piece of feedback that has been sent in from your customers—whether a text message, picture message, voice recording, web form, or email, usually contains a score and a comment to explain the rating. It may not always contain a comment; it could be a customer satisfaction rating or other metric. For more information, see Items of feedback.

The icon next to the feedback indicates the type of message:

SMS text message SMS text message
Voice recording Voice recording
Email message Email message
Twitter Facebook Social media post
Web or web widget Web or web widget message

Each piece of feedback is then processed by the Rant & Rave Sentiment engine, which is configured during project setup to identify those topics that are relevant to your organisation. The Sentiment engine then creates relationships between these responses and the topics to create insights that effectively reveal how your customers feel about most important aspects of your organisation’s offerings.

The Sentiment engine is only interested in one bit of the feedback—the comment, which it processes to understand:

  • What is the customer talking about (TOPICS)
  • How do they feel (SENTIMENT)
  • Can we match that feeling to what they are talking about (INSIGHT)

Let’s take a look at the following example comment:

“Your agent was so helpful and friendly but the website was confusing.”

As you can see from this example, a comment can have multiple topics (underlined) and very different insights (in green and red).

This is why the Sentiment engine is so important – feedback is so much more than just a simple score!

Basic terminology

Sentiment score – The Sentiment engine works out an overall score for the emotion in the comment. The score is based on a scale of 1 to 5 (1 = negative, 5 = positive) and defaults to a neutral 3.

Categories - Categorisation is the process of putting the topics into categories. A comment can be categorised more than once. The comment above sits in two buckets: website and service. We call our categorisation structure a taxonomy because it contains a hierarchy; that is, categories can contain subcategories.

Customer-given score – This is the score that a customer gave when they completed their Fast Feedback survey. The score is a number, but the scale depends on the client. It could be 1 to 5 or another range, for example, 0 to 10 for NPS. Fast Feedback captures the score but does not change it.

Comment – These are the actual words that the customer left as feedback. They might have sent the comment by SMS, via the web widget, or on one of our IVRs (interactive voice responses) as a recorded comment. The comment is what the sentiment engine processes but does not change the comment itself.

Item – A line of data in Fast Feedback is called an item. Usually this is a customer-given score and a comment, sometimes with additional structured fields.

Structured fields/metadata – This is data that we get from the client. It includes things like the agent name, team leader, which branch a customer was in, what the customer ordered etc. These fields are defined as part of the project to set up the service. They are used to help analyse the feedback and also to control who sees what in the Dashboard (for example, a team leader can only see feedback about their team).