Key Terms and Metrics

There are several key terms and metrics that you will need to understand to use Interaction Analytics (IA) successfully. To quickly find specific information on this page, press Ctrl+F on your computer keyboard. In the resulting text box, type the information you are looking for. You can also look for the term or metric you would like to understand better in the navigation menu to the right.

Categories

A way to filter interaction data into groups or topics that you want to measure and track. They make it easier to find trends in your data. You can add custom categories to those that come out-of-the-box with Interaction Analytics on the Language Model page.

Category Folder

Groups of related categories for easier reference. On the Language Model page, you can group related categories in folders and sub-folders. This lets you find them more easily.

Category Set

Sets of related categories to be displayed on a page or in a widget. Category sets let you create custom groups of categories. Interaction Analytics comes with a variety of categories out-of-the-box. It also includes 13 category templates, such as Intent to Buy and Risk Aversion. You can import and modify the default templates. You can also create your own custom templates.

Custom Sentiment

You can assign a custom sentiment to certain keywords, phrases, and custom entities in your company profile. For example, in Interaction Analytics, the word cancel is neutral by default, but you may want it to be negative if you're concerned about customers canceling a service. See the sentiment entry on this page to learn more about sentiment.

Datasets

A named set of interaction transcripts that span a specific time period and match any criteria you dictate. When you create a new dataset, it includes all analyzed interactions by default. You can configure datasets to only include interactions that meet specific criteria. You can filter the dataset to include or exclude interactions using certain keywords, phrases, or entities. Other filters include sentiment, metrics, frustration, and resolution status.

Using a single dataset that includes all interactions is usually best. However, you may want to separate datasets by purpose or business unit. For example, you may want to separate datasets by language. Or if you have multiple teams of data analysts, you may want to create a dataset for each team.

Entities

Specific themes, words, or concepts in your data. Interaction Analytics contains some entities out of the box. For example, the "sound quality" entity applies to transcripts where people say key phrases like I can't hear you or you're breaking up. If You can also create custom entities for specific entity types in the company profile.

Filters

Tools you can use to view more specific data on the Search page and in workspace widgets. You can filter by teams, channels, participant, and other metrics like sentiment. The criteria set for a filter can be saved as a search or category.

On the Search page, you can use filters to search across all your data without being restricted by datasetsClosed Set of parsed interaction transcripts that spans a specific period of time and match your specified filter criteria. or categoriesClosed Data groups that Interaction Analytics uses for parsed call transcript data. They make it easier to find trends.. You can save specific search parameters to apply to workspaces. Within workspaces, you can apply filters to only view data from certain interactions in a dataset. Workspace do not impact the data that displays in Interaction Analytics prebuilt reports.

Frustration

Frustration cues are words and phrases that indicate customer frustration based on language in an interaction transcript. Frustration cues are not influenced by voice characteristics, such as tone, volume, and speed.

Note that negative sentiment is different from frustration. All frustrated interactions should also be negative, but in rare cases negative sentiment may not correlate with frustration. Some phrases can be both negative and frustrated. However, a client may be discussing a negative issue they want addressed without necessarily being frustrated. Frustration only measures cues for the client, while sentiment can measure cues for both the client and the agent. Frustration is a better filter to use to investigate why clients are upset, disappointed, or angry.

See the sentiment entry on this page to learn more about negative sentiment.

Interaction

A conversation between an agent and a client. An interaction can consist of multiple contacts if the same client has multiple complete conversations regarding the same issue. A contact can also consist of multiple segments if a client is transferred to a different agent or skillClosed Used to automate delivery of interactions based on agent skills, abilities, and knowledge. IA stores recordings and transcriptions for all interactions your organization inputs. This can include voice calls or chats from any channel you use.

Keywords and Phrases

Specific words or phrases with significance to your company. Interaction Analytics can tell you the top keywords and phrases occurring in your interactions. You can also include keywords and phrases in dataset criteria, categories, and use them to filter data.

Metrics

Metrics are attributes and measurements for different interaction events and data. You can use metrics to filter datasets, workspaces, and widgets. You can also use them to filter searches across all your available data. Each metric requires different configuration, such as entering a value or selecting an integer. For example, you can use the Team Name metric to only view interactions for a specific team. Once selected, you need to enter the name of the agent you would like to see data for. Not all the same metrics are available in all areas of IA.

Notable Silence

When neither the client nor the agent are talking during an interaction. Silence is considered notable if it lasts three or more seconds. Total notable silence for an interaction is the sum of all notable silence periods. Chats tend to have a higher silence percentage voice interactions because it may take longer for a client or agent to type a response.

Resolution Status

Interaction Analytics assesses both sides of an interaction to determine whether the contact reason is resolved or unresolved. It looks at the whole conversation for questions and sees if they were answered. It also searches for indications that the customer is not going to be helped.

You can view resolution status data in the Most Unresolved Interactions widget, Interactions widget, and Interaction Analytics BI reports.

Resolved

An interaction is resolved if the contact received satisfactory help and answers to their questions. Resolved interactions usually don't require further action, such as a transfer or callback. However, not all transfers or callbacks mean an interaction should be unresolved. For example, if an agent clarifies that a transfer is the appropriate next step or the client asks for it, the interaction is still resolved.

Unresolved

An interaction is unresolved if the contact did not receive satisfactory help or answers to all their questions. Unresolved interactions usually indicate a need for additional action, such as a callback, transfer, or reschedule. But, not all transfers or callbacks mean an interaction is unresolved.

The Most Unresolved Interactions widget displays unresolved contacts by agent or category. It also shows the talk time for each respective agent or category.

Sentiment

Client and the agent cues are evaluated for positive or negative indicators and strength. Sentiment cues are words and phrases that indicate mood based on language in an interaction transcript. Sentiment scores are not influenced by voice characteristics, such as tone, volume, and speed. These cues indicate the sentiment of an interaction. Sentiment scores in Interaction Analytics can be positive, negative, mixed, or neutral.

Sentiment is color-coded consistently so that any sentiment data in widgets or charts is represented by the following color assignment:

  • Negative: Red

  • Positive: Blue

  • Mixed: Dark gray

  • Neutral: Light gray

You can choose to analyze overall sentiment, beginning sentiment, or end sentiment.

  • Overall Sentiment: The sentiment for the whole interaction.
  • Beginning Sentiment: The sentiment at the beginning of an interaction. This is determined by the first 400 words, or first 30%, of the interaction, whichever occurs first. It can help you identify common reasons or problems clients call about. It can also help you compare against end sentiment. If the client or agent sentiment changed over the course of an interaction, you can explore why. Mixed sentiment is not available for beginning sentiment.

  • End Sentiment: The sentiment at the end of an interaction. End sentiment is determined by the last 30% of an interaction. It can help you compare against beginning sentiment. If the client or agent sentiment changed over the course of an interaction, you can explore why. Mixed sentiment is not available for end sentiment.

Negative Sentiment

The sentiment score for an interaction is negative if the number of negative cues is greater than the number of positive or neutral sentiment cues. Negative sentiment indicates a negative event or a potential problem. For example, some negative sentiment cues are, It's not working or I'm having trouble hearing you.

Negative sentiment doesn’t have to mean there’s a problem with agent performance or behavior. It can also stem from a product issue, a billing challenge, and other events outside agent control.

Note that negative sentiment is different from frustrationClosed Looks for cues to identify customer frustration. The cues include words and phrases like, "I'm very angry". Frustration isn't the same as negative sentiment. Frustration cues show that a customer is upset rather than just discussing something negative.. Frustration is a better filter to use to investigate why clients are upset, disappointed, or angry. See the entry for frustration on this page to learn more about that metric.

Positive Sentiment

The sentiment score for an interaction is positive if the number of positive cues is greater than the number of negative or neutral cues in an interaction. Positive sentiment cues could include phrases such as That's great! or I'm interested. For non-voice interactions happy emoticons are also be considered positive cues.

Mixed Sentiment

Mixed sentiment occurs when an interaction contains a similar amount of both positive and negative cues. For example, I didn't get my order. Thanks for your help. would result in mixed sentiment.

Neutral Sentiment

Sentiment is neutral if there is an absence of negative or positive cues.

You can assign a custom sentiment to certain keywords, phrases, and custom entitiesClosed Keyword or phrase defined in your company profile in Interaction Analytics. Related to an entity type. Can include variants. in your company profileClosed Page in Interaction Analytics where you can create custom entities.. For example, in Interaction Analytics, the word cancel is neutral by default, but you may want it to be negative if you're concerned about customers canceling a service. See the custom sentiment entry on this page to learn more or you can configure custom sentiment in you company profile.

Transcript

Written form of interactions. Interaction Analytics collects chat interactions immediately, since they're already in written form. Voice interactions must be transcribed. They're collected by Interaction Analytics as soon as transcription is complete. Transcripts are then parsed. This means they are broken into parts to identify keywords, phrases, and entities.

Widgets

Widgets appear inside workspaces in Interaction Analytics. They are graphical representations of data that meet specified filter criteria. For example they may include bar graphs or bubble charts to show trends, percentages, or common words found in interaction data. You can edit, filter, add, and delete widgets from a workspace.

Workspace

Workspaces are collections of one or more related widgets displayed in a specific way. When a user opens a dataset, they see their workspace, like a dashboard, for that dataset. This gives them the information they need to make decisions.

Interaction Analytics comes with a variety of helpful workspaces out of the box. You can also create custom workspaces.

Workspace Template (Set)

Workspace templates, or sets, are groups of workspaces that usually relate to a specific theme. Interaction Analytics comes with a variety of helpful workspace templates out-of-the-box. You can also create custom templates for your unique needs. You can apply a workspace template to a dataset so that users opening that dataset will have all the information they need at their fingertips.