Datasets are sets of interactions from your contact center that have been recorded and parsed
Process of analyzing data and breaking it into parts. for you to analyze. These interactions include voice calls, emails, chats, supported digital channels, or any combination of these. Voice calls include written transcripts you can view. Datasets span a specific time period and match any filter criteria you apply.
You can use filter criteria to include or exclude interactions that meet some or all of the qualifications you enter. For example, you could create a dataset of interactions spanning the last 90 days that include a keyword or fall into a specific category.
While you can filter datasets to only include interactions that meet certain criteria, it's best to use filters at the workspace level instead. You may want to create separate datasets for different languages or for each customer or account. However, it's not recommended to create many datasets or heavily filtered datasets. This makes managing your data in IA less complicated and avoids excluding interactions with potentially valuable data.
Clarissa Dalloway is an administrator at Classics, Inc. Classics has customers in different regions that speak English, Spanish, and French. Clarissa creates three separate datasets for the interactions in each of these languages. She creates clear and distinct names for each dataset and configures each with a Dataset Period of the Last 90 days. She configures the desired channels for each dataset and clicks Save. She chooses not to Add Criteria to filter the dataset, so that all interactions in the designated languages are included. That way, she can use workspace and widget filters to sort through the data in those interactions.
Classics has a designated team of agents that handle escalated interactions. So, the Classics administrator, Clarissa Dalloway, decides to create a dataset just for the Escalations team. She adds a Metric filter to the dataset criteria and selects the teamName metric. Then she selects the name of the Escalations team.
While she could have created a dataset without any criteria and filtered a workspace for this team instead, she decided this would be more helpful to focus on issues causing escalations.
Key Facts About Datasets
- Voice data becomes available as soon as the call is transcribed into text. This typically takes three to five hours after the call ends, depending on the length of the call and the call volume at the time. When call volume is low, data may become available within as little as an hour. However, it may take several hours if call volume is high.
- Email, chat, and digital interactions are available immediately after the interaction ends.
- Up to 15 datasets are supported per business unit High-level organizational grouping used to manage technical support, billing, and global settings for your CXone environment.
- You cannot change a dataset name after you create it. You cannot use special characters in the dataset name. This makes it important to carefully plan naming conventions for your datasets in advance. Make sure you can quickly and easily identify the data contained in each dataset. It can also be helpful to know at a glance who created the data set. Some users find that the dataset topic or name of the team being analyzed followed by the creator initials is a useful naming convention. In the previous example, Maria could create a dataset named Fables Mentions MB.
- If your company has Interaction Analytics for multiple languages, you can create datasets for the different languages. You cannot create a single dataset that includes interactions in more than one language.
- You can filter the data contained in a dataset based on keywords, phrases Combination of words that have special significance when used together in a specified order, such as "want to cancel"., or entities Keyword or phrase defined in your company profile in Interaction Analytics. Related to an entity type. Can include variants. interactions contain or do not contain. You can choose which channels to include and the period of time to cover. You can include or exclude interactions by sentiment Overall mood or result of the interaction as determined by analysis of words, phrases, and context of the transcript., frustration 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., or resolution. You can also filter by metrics tags like specific teams, skills, or agent names. The maximum number of metric tags you can add to filter criteria is 50.
- When you create a dataset, you can apply a workspace template Multiple workspaces grouped as a reusable template. right from the dataset page.