Topics

Topics capture customer feedback about specific business concerns

Medallia Text Analytics uses topics to capture words and phrases from text feedback. The following table describes how topics capture feedback based on keywords, attributes, and experiences.

Usage

Definition

Examples

Time estimates

Keyword

Retrieves keywords about a particular subject. Rules for keyword topics are typically a series of synonyms.

  • Hotel Shuttle

  • Breakfast Offerings

  • Fitting room
  • Build: ~30-60 minutes
  • QA: ~10-15 minutes

Attribute

Retrieves comments about an attribute or characteristic of a certain subject.

  • Staff Knowledge

  • Bathroom Cleanliness

  • Product - Quality
  • Build: ~1.5-2 hours
  • QA: ~15-30 minutes

Experience

Retrieves comments about customer experiences with aspects of the company. Experience topics are often about:

  • the ease or difficulty of doing something

  • the experience of locating or finding something

  • the time it takes to do something

  • Website Ease of Use

  • Ease of check-in

  • Locating the hotel
  • Build: ~2-3 hours
  • QA: ~20-30 minutes
Note: Avoid creating topics that are tinged by positive or negative sentiment (such as Attitude–Good and Attitude–Bad). Such topics always capture predominantly positive or negative comments, making it difficult to gauge improvement, and making the impact score less helpful.

Topic types

There are two types of topic: standard and compound. Standard topics find keywords in customer comments, such as in the text fields of surveys. For example, you might use a standard topic to identify when comments mention unfriendly staff or dirty restrooms.

Compound topics include other topics in rules, and can identify when phrases were mentioned specifically by the customer or the agent in conversational data. For example, you might want to identify interactions where the agent used appropriate empathy after the customer asked to close their account. In this case, you might create a compound topic with a rule that finds customer-mentioned phrases tagged by the Call Reason - Close Account topic followed by agent-mentioned phrases tagged by the Agent - Empathy topic.

Note: Compound topic rules can include only child standard topics. You cannot use parent standard topics or other compound topics in compound topic rules.

You must select a topic type as part of the topic-creation process.

Topic rules

A topic rule is the combination of logic and words (including keywords, nearby words, and linguistic connections) that define which phrases are associated with a particular topic. Each captured phrase is associated with a particular rule; add or modify rules to include or exclude phrases. For more information about developing topic rules, see Build topics. Each topic type has a different process for creating rules, as described in Create standard topic rules and Create compound topic rules.

Topic hierarchies

Another factor to consider when creating and using topics is the topic hierarchy. Topics are usually organized in hierarchies of parent topics and child topics.

A parent topic is a topic with one or more child topics. Parent topics inherit comments from child topics, but should not contain rules inside them. These topics help you gauge customer opinion about more general subjects (such as Website).

A child topic is a topic that belongs to a parent topic. Child topics help you gather customer comments about more granular subjects (such as Website - Navigation and Ease of use and Website - Performance).

The following figure shows an example of a parent-child topic hierarchy:

A parent topic called Website, with child topics Website - Navigation and Ease of use, Website - Performance, and Website - Login and PW