Investigate Your Data in AutoDiscovery

Use the AutoDiscovery map and metrics to investigate the suggested topics and phrases in your dataset. You can search, filter, and sort your data to identify trends and anomalies that may be of interest to you. The information you discover can help you edit or create new categories for your dataset. You can export your findings as a CSV file to review and share later.

Investigate Suggested Topics on the Map

To investigate the map you can:

Zoom in and out or pan over the map to view topics and phrases.
Click on a topic that you are interested in. The map focuses on the topic and the topic statistics display in the Statistics panel.
Click into the topic on the map to view the phrases included in a topic. Proximity to the center of a topic indicates the frequency of that phrase within the topic.

Filter Your Data

You can filter your data by specific metrics or topics and phrases. There are a variety of tools you can use to focus on areas of interest quickly and easily. AutoDiscovery processes a maximum of 50,000 interactions at once so these filters are helpful to narrow your analysis to interactions you are most interested in.

Learn more about filter options

Filter	Details
Channels	Filters data by channels used by your organization. Supported channels include: Inbound and outbound Voice Omnichannel and Digital Experience Chat Omnichannel and Digital Experience Email SMS. Digital Experience channels: Apple Business Chat Facebook Messenger Google Business Messages In-App Messaging In-App Messaging LINE Messaging Microsoft Teams Slack Twitter Direct Messages Telegram Messaging Viber WhatsApp Custom channels using the Bring Your Own Channel feature
Frustration	Filters data based on whether the contact was frustrated. Select one of the following Frustration options: Frustrated: Only displays interactions where the client (contact) was frustrated. Not Frustrated: Only displays interactions where the client was not frustrated. Note that frustration is different from negative sentiment. All frustrated interactions should also be negative, but in some cases negative sentiment could mean something else. Frustration cues show that a contact is upset. For example, frustration cues could include words or phrases like , "I'm very angry."
Readability	Filters data based on how easy it is to read the interaction transcript. Automatic Speech Recognition (ASR) assigns confidence ratings to transcripts based on the number of errors they have. The fewer the transcript errors, the easier it is to read. So, this is helpful if you have to read many transcripts to solve an issue and want to focus on transcripts that are easier to read. By default, all transcripts are included. Select one of the following Readability options for an ASR confidence rating: High: Only displays interactions with transcripts that are easier to read. Low: Only displays interactions with transcripts that are more difficult to read. Important: Filtering by High confidence will likely still return transcripts with errors, but they will be easier to read than lower quality transcripts.
Resolution	Filters data based on whether the cause for the interaction was resolved during the interaction. Select one of the following Resolution options: Resolved: Only displays data for resolved interactions. Unresolved: Only displays data for unresolved interactions.
Keywords and Phrases	Includes or excludes interactions from the data based on keywords or phrases they contain. You can enter a keyword or phrase to return exact matches or variations. For phrases, you can also specify how close the words in the phrase must be to each other for an interaction to be considered a phrase match. You can choose to view data from only the agent side, only the contact side, or both sides of interactions. If you only filter for one side of interactions, you can still see both sides when you drill down to the transcript view. Select one of the following options: Keywords and Phrases - Agent: Only displays agent data. Keywords and Phrases - Client: Only displays client (contact) data. Keywords and Phrases: Displays both agent and client data. Enter keywords or phrases you want to filter for. Select one of the following Keywords and Phrases options from the first drop-down: Have: Displays data for interactions that mention the word or phrase you enter. Have at least one: Displays data for interactions that mention at least one of the terms or phrases you enter. This is only helpful if you use more than one keyword or phrase filter. Not have: Only displays data for interactions that don't mention the term you enter. Select one of the following options from the Matching drop-down: Exact: Only displays interactions that include the term or phrase exactly as you enter it. Approximate: Displays interactions that include variations of the term or phrase you enter. For example, if you enter activate, it could display interactions that include activate, activates, activating, activated, and so on. If you entered a phrase, use + and - or enter an integer to configure Proximity. This includes matches where the words in the phrase aren't consecutive. Phrase matches can have up to eight words between them. The default is 0. For example, the phrase speak manager with a proximity of 2 would capture the phrases speak to a manager, speaker with your manager, and so on. Click Add when you're finished. Repeat these steps to add additional filters. If you add two or more of this filter, there is an “AND” relationship between words. This means that each item you add to the filter must be present in the interaction to display as a filter result. For example, if you add a Keywords and Phrases - Agent filter for the word awesome, a Keywords and Phrases - Client filter for great, and a filter for when either participant says thank you, all of these must be said for the filter to find the interaction in this case.
Categories	Filters data by existing category rules. You can choose to view data from only the agent side, only the contact side, or both sides of interactions. If you only filter for one side of interactions, you can still see both sides when you drill down to the transcript view. Select one of the following options: Categories - Agent: Only displays agent data. Categories - Client: Only displays client (contact) data. Categories: Displays both agent and client data. Drill down and select the categories you want to filter for. Selected categories become blue. If you add two or more of this filter, there is an “AND” relationship between words. This means that each item you add to the filter must be present in the interaction to display as a filter result.
Entities	Includes or excludes interactions from the data based on entities they contain. Some entities are preconfigured in Interaction Analytics. If you want to create custom entities, you must configure them in the company profile. If you want to filter based on keywords and phrases, use that filter instead. You can choose to view data from only the agent side, only the contact side, or both sides of interactions. If you only filter for one side of interactions, you can still see both sides when you drill down to the transcript view. Select one of the following options: Agent: Only displays agent data. Client: Only displays client (contact) data. Either: Displays both agent and client data. Select whether you want to Include Entities or Exclude Entities, then choose the entities you want to include or exclude. If you add two or more of this filter, there is an “AND” relationship between words. This means that each item you add to the filter must be present in the interaction to display as a filter result.
Sentiment	Filters data by sentiment. You can choose to view data from only the agent side, only the contact side, or both sides of interactions. If you only filter for one side of interactions, you can still see both sides when you drill down to the transcript view. Select one of the following options: Agent: Only displays agent data. Client: Only displays client (contact) data. Either: Displays both agent and client data. Select one of the following sentiment types: Overall Sentiment: The sentiment for the whole interaction. Beginning Sentiment: This is determined by the first 400 words, or first 30%, of the interaction, whichever occurs first. It can help you identify common reasons clients call about. Mixed sentiment is not available for beginning sentiment. End Sentiment: End sentiment is determined by the last 30% of an interaction. Mixed sentiment is not available for end sentiment. Once, you've selected a sentiment type, choose one or more of the following sentiments: Positive: The number of positive cues is greater than the number of negative or neutral cues in an interaction. Negative: The number of negative cues is greater than the number of positive or neutral sentiment cues in an interaction. Note that negative sentiment is different from frustration. Some phrases can be both negative and frustrated. However, a client (contact) may be discussing a negative issue they want addressed without necessarily being frustrated. Neutral: An absence of negative or positive cues. May also occur if you have phrases set as neutral in your company profile. Mixed: The number of positive and negative cues in an interaction are essentially equal. If you don't select a sentiment the default is to include interactions of all sentiments. If you add two or more of this filter, there is an “AND” relationship between words. This means that each item you add to the filter must be present in the interaction to display as a filter result.
Company Profile	Filters for entities configured in your company profile. You can filter by the entity type, like Products, or drill down to include entities within the entity type, like Sweaters. The options will vary depending on the entities and entity types you have added to your company profile. You can choose to view data from only the agent side, only the contact side, or both sides of interactions. If you only filter for one side of interactions, you can still see both sides when you drill down to the transcript view. Select one of the following options: Agent: Only displays agent data. Client: Only displays client (contact) data. Either: Displays both agent and client data. To include all configured entities under a single entity type, such as Company, select the checkbox beside the entity type name. Clear the checkbox to remove it. To select only certain entities under an entity type, click the filter you just selected again and select the checkbox next to each entity you want to include. Clear the checkbox to remove it. If you add two or more of this filter, there is an “AND” relationship between words. This means that each item you add to the filter must be present in the interaction to display as a filter result.
Metrics	Check the box next to the specific metric type you would like to filter for. After you check the box for the metric you want, click the filter you just added to select a value for the specified data field. Click the check mark to the right when you are finished. For example, if you select Agent Name as a metric, you would need to enter or select the name of the agent you would like to filter for. See the Metrics entry in Key Terms and Metrics for more information about each available metric type.
Customer Satisfaction	This filter is only visible if your system enables CXone Mpower AI features. It filters for specific CXone Mpower metrics and the scores for those metrics. These metrics and scores measure behaviors that demonstrate efforts to achieve customer satisfaction. Check the box next to the customer satisfaction behaviors you want to include in your search. Then select one or more scoring level for the metrics you want to see. The scoring levels include: Moderately Negative, Neutral, Moderately Positive, Strongly Positive.
Sales Effectiveness	This filter is only visible if your system enables CXone Mpower AI features. It filters for specific CXone Mpower metrics and the scores for those metrics. These metrics and scores measure behaviors that demonstrate efforts to make sales effectively. Check the box next to the sales effectiveness behaviors you want to include in your search. Then select one or more scoring level for the metrics you want to see. The scoring levels include: Strongly Negative, Moderately Negative, Neutral, Moderately Positive, Strongly Positive.
Intents	Filters by the intent of the interaction as assigned by CXone Mpower. This allows you to view interactions by purpose based on what the contact wants to communicate or accomplish. Only one intent is assigned to an interaction, but you can filter by multiple intents at once. Intents are sorted into three different levels: category, topic, and intent. If you select the box next to a category, it also filters by the topics and intents included in that category. Intents are organized by topic, and topics are organized by category. If you only want to filter by certain intents, click the drop-downs of the relevant category and topic to drill down to the desired intent. It may be helpful to create categories by intent. This is supported for voice and digital channels. Supported digital channels include all conversational digital interactions but not social media posts. They also include legacy ACD chat and SMS channels. Legacy ACD email and CXone Mpower email are not supported. To use this, you must have one of these licenses or license combinations: Analytics Advanced Analytics Premium Analytics Essential and AutoSummary If you have questions, ask your Account Representative.
Outcomes	Filters by the outcome of an interaction. For example, outcomes could include payment completion, a transfer, or an escalation. Interactions can have up to 12 outcomes. Select the box next to the outcomes you want to filter for. It may be helpful to create categories by outcome. This is supported for voice and digital channels. Supported digital channels include all conversational digital interactions but not social media posts. They also include legacy ACD chat and SMS channels. Legacy ACD email and CXone Mpower email are not supported. To use this, you must have one of these licenses or license combinations: Analytics Advanced Analytics Premium Analytics Essential and AutoSummary If you have questions, ask your Account Representative.
Actions	Filters by actions that occurred during an interaction. For example, actions could include processing a payment, providing account information, or pulling up an account. Interactions can have up to eight actions. Select the box next to the outcomes you want to filter for. It may be helpful to create categories by action. This is supported for voice and digital channels. Supported digital channels include all conversational digital interactions but not social media posts. They also include legacy ACD chat and SMS channels. Legacy ACD email and CXone Mpower email are not supported. To use this, you must have one of these licenses or license combinations: Analytics Advanced Analytics Premium Analytics Essential and AutoSummary If you have questions, ask your Account Representative.

Filter Using the Data Layer

You can filter the topics in the dataset using one or more metrics in the Data Layer panel. Metric sliders allow you to focus on more specific metric results. The data displayed on the map and the Statistics panel will change as you make adjustments in the Data Layer panel. Note that the initial view of the map is set to the Volume metric.

Clarissa Dalloway is a contact center administrator at Classics Inc. She wants to see if there is a connection between customer satisfaction and how long customers spend on the phone with agents. She opens the Data Layer panel in AutoDiscovery and selects the Sentiment metric. She moves the metric slider to the right to only see results with negative sentiments. Next, she selects Duration from the Size drop-down. The color of the circles on the map are orange and red to show topics associated with negative sentiment. The size of the circles show her how long calls lasted. The bigger the circle, the longer the call length. Now she can look at specific topics and interactions to find out more.

Click the app selector and select Analytics.
Click AutoDiscovery.
Click to open the Data Layer panel.

Select one or more metrics you would like to use. The topics are shown in a gradient of the metric colors on the map. You can use the metric sliders to focus on more specific results for each metric type. As you move the slider, topics are filtered on the map and in the Statistics panel. Topics that do not match the filter move into the background on the map.

Learn more about fields in this step

metric	description
Volume	Total number of calls using the selected topic or phrase. The Statistics panel presents a graph with the volume data for the selected time period. You can click on points in the graph to see the volume per day. Volume is represented by circles ranging from light to dark blue on the map. The darker the shade of the blue, the higher the volume of calls for that topic.
Duration	Length of a call between an agent and a customer. This can highlight specific issues that agents may find difficult to deal with or solve. Duration is represented by circles ranging from light to dark gray on the map. The darker the shade of gray, the longer the duration of calls.
Anomaly	Instances where there is an unusual trend in one of the metrics. There are likely unexpected changes or spikes in the data. This can help you to identify and resolve problems. Anomalies are represented by circles ranging from light to dark purple on the map. The darker the shade of purple, the higher the anomaly score.
Silence	Periods of silence where there is no conversation between customer and agent. This may indicate the agent is struggling to deal with the problem or is unsure how to proceed. Circles on the map will display in a gradient of shades of green, yellow, orange, and red. Dark green indicates a low percentage of silence. Red indicates a high percentage of silence. Colors ranging in between indicate intermediate percentages of silence.
Cross Talk	Periods where both customer and agent are speaking at the same time. Cross-talk may indicate conflict between customer and agent, or it can signify noncompliance of company policy regarding customer contact etiquette. Circles on the map will display in a gradient of shades of green, yellow, orange, and red. Dark green indicates where there were short periods of cross talk. Red indicates long periods of cross talk. Colors ranging in between indicate intermediate amounts of cross-talk.
Customer %	Percentage of time the customer spoke during the call. Circles on the map will display in a gradient of shades of green, yellow, orange, and red. Dark green indicates that customers spoke for a small percentage of calls. Red indicates that customers spoke more during calls. Colors ranging in between indicate that the customer spoke for an intermediate percentage of the call.
Sentiment	Measures customer satisfaction and sentiment over the entire call. Sentiment is measured using AI and machine learning technology. Circles on the map will display in a gradient of shades of green, yellow, orange, and red. Dark green indicates positive sentiment. Red indicates negative sentiment. Colors ranging in between indicate neutral or mixed sentiments.
Negative %	Measures the percentage of calls where sentiment is below a certain threshold. This threshold signifies the difference between customer satisfaction and negative sentiment. Circles on the map will display in a gradient of shades of green, yellow, orange, and red. Dark green indicates low negative sentiment. Red indicates high negative sentiment. Colors ranging in between indicate intermediate negative sentiment.

To focus on only one metric, click the Size drop-down and select Same as color. The circle size and color will both relate to the selected metric.
To focus on more than one metric at a time, click the Size drop-down and select the metric you would like the circle size to represent on the map. The color of circles on the map will reflect the selections you made with the metric sliders. The size of circles will illustrate the prevalence of the metric you chose in the drop-down.

You can identify a topic you want to investigate further and click to drill down to see the interactions for the topic.

You can click Show all in the Data Layer panel to reset the map.

Filter Using the Statistics Panel

The Statistics panel lists all the topics and phrases within the dataset. You can search for specific topics or phrases using the search bar or filter by metric.

Click the app selector and select Analytics.
Go to AutoDiscovery.
Click the arrow to the left to open the Statistics panel.
Click the Topics drop-down to view either Topics or Phrases. You can click the arrows to the left to sort the list in alphabetical order.

Select a metric from the drop-down. You can click the arrows to the left to sort the list in ascending or descending order.

Learn more about fields in this step

To focus on a specific topic or phrase, enter it in the search bar or select one from the list in the Statistics panel. The statistics for each of the metrics displays for the topic or phrase you selected.

View image
To view all the phrases within a topic click .
Click the phrase you would like to view. The map zooms in and the panel displays statistics for that phrase.

metric	description
Volume	Total number of calls using the selected topic or phrase. The Statistics panel presents a graph with the volume data for the selected time period. You can click on points in the graph to see the volume per day. Volume is represented by circles ranging from light to dark blue on the map. The darker the shade of the blue, the higher the volume of calls for that topic.
Duration	Length of a call between an agent and a customer. This can highlight specific issues that agents may find difficult to deal with or solve. Duration is represented by circles ranging from light to dark gray on the map. The darker the shade of gray, the longer the duration of calls.
Anomaly	Instances where there is an unusual trend in one of the metrics. There are likely unexpected changes or spikes in the data. This can help you to identify and resolve problems. Anomalies are represented by circles ranging from light to dark purple on the map. The darker the shade of purple, the higher the anomaly score.
Silence	Periods of silence where there is no conversation between customer and agent. This may indicate the agent is struggling to deal with the problem or is unsure how to proceed. Circles on the map will display in a gradient of shades of green, yellow, orange, and red. Dark green indicates a low percentage of silence. Red indicates a high percentage of silence. Colors ranging in between indicate intermediate percentages of silence.
Cross Talk	Periods where both customer and agent are speaking at the same time. Cross-talk may indicate conflict between customer and agent, or it can signify noncompliance of company policy regarding customer contact etiquette. Circles on the map will display in a gradient of shades of green, yellow, orange, and red. Dark green indicates where there were short periods of cross talk. Red indicates long periods of cross talk. Colors ranging in between indicate intermediate amounts of cross-talk.
Customer %	Percentage of time the customer spoke during the call. Circles on the map will display in a gradient of shades of green, yellow, orange, and red. Dark green indicates that customers spoke for a small percentage of calls. Red indicates that customers spoke more during calls. Colors ranging in between indicate that the customer spoke for an intermediate percentage of the call.
Sentiment	Measures customer satisfaction and sentiment over the entire call. Sentiment is measured using AI and machine learning technology. Circles on the map will display in a gradient of shades of green, yellow, orange, and red. Dark green indicates positive sentiment. Red indicates negative sentiment. Colors ranging in between indicate neutral or mixed sentiments.
Negative %	Measures the percentage of calls where sentiment is below a certain threshold. This threshold signifies the difference between customer satisfaction and negative sentiment. Circles on the map will display in a gradient of shades of green, yellow, orange, and red. Dark green indicates low negative sentiment. Red indicates high negative sentiment. Colors ranging in between indicate intermediate negative sentiment.

You can identify a topic you want to investigate further and click to drill down to see the interactions for the topic.

Investigate Anomalies and Trends

AutoDiscovery identifies unusual occurrences in your dataset. Anomalies show changes in trends over time. This highlights topics and phrases that may need your attention.

Classics Inc. is offering customers a new membership rewards program. The details about the membership on the Classics Inc. website are unclear, so they are receiving an increased amount of calls about it. Anomalies in AutoDiscovery display common topics and phrases related to the rewards program. Analyzing these topics and phrases can help Classics Inc. identify how to make their website more informative and improve customer satisfaction.

Identify Anomalies with the Statistics Panel

Open the Statistics panel.
Select either Topics or Phrases from the drop-down.
Select Anomalies from the metric drop-down.
Sort the list in descending order, from highest to lowest anomaly score.
Select a topic or phrase to view statistics and the trend graph.
Move your mouse over the graph to view details about the peaks.
If you selected a topic, you can drill down further and do the same to focus on the topics' phrases.

Identify Anomalies with the Data Layer

Open the Data Layer panel.
Select the Anomalies metric.
Move the metric slider to the right to see to focus on areas with high anomalies.
To only see results related to anomalies, click the Size drop-down and select Same as color.
To identify other factors contributing to the anomaly, try selecting other metrics.
Select a specific topic or phrase in the topic to focus on it.

View Interactions

After finding a topic or phrase you want to focus on, you can click to view individual interactions.

Once you open the Interactions Widget you can:

Sort the columns in ascending or descending order.
Select an interaction to see the corresponding transcript and listen to the call. Keywords are highlighted in the transcript.
Export data to a CSV file.

Add Phrases to a Category

AutoDiscovery can help you identify new categories you may want to create or phrases you may want to add to an existing category.

In AutoDiscovery, select the topic or phrase you want to add to a category and click .
In the Select Location window, expand the Categories list and select a folder to create a new category, or select an existing category to add the phrase to it and click Save as Category.
Use the Rules Editor to make any desired changes to the category rules. This allows you to select who you want rules to apply to such as the agent, client, or neither. You can also modify existing rule sets by adding keywords, phrases, sentiment, metrics, company profile entities, or other elements to a category.

If you make any updates in the category editor, the dataset will be different from what was previewed in your workspace.

Once you have made your changes, you can navigate back to AutoDiscovery using the link at the top on the Rules Editor window.

Save a Search

You can select saved searches from the drop-down to view search results again. This helps you avoid having to configure the same search settings multiple times. You can save a search as public or private. Public searches can be seen and applied by other users. Private searches can only be seen and used by the user that created them. In AutoDiscovery you can only save date range or filter settings applied from the drop-down. Saved searches don't include any filters or settings specific to AutoDiscovery.

It's helpful to develop a consistent naming convention for searches, so you can quickly identify the one you want to apply. Selecting a unique color for each new saved search also makes them easier to distinguish.

Click the app selector and select Analytics.
Go to AutoDiscovery.
To save a search, configure the search settings and filters for the data you would like to view and enter your search.
When you are finished, click the save icon in the top right corner.
In the New Search field, enter a name for your search. You cannot use the following characters in the name of a saved search: [ ] { } < >.
To select a different color for your search, click the colored box in the New Search field and select another color.
Enter a description of your search settings and filters.
Enable Public, to make the search visible to other users. Saved searches are private by default.
Click Save As. Now you will be able to select the same search criteria from the drop-down in the future.

You can view details, copy, and delete saved searches by clicking more options icon of 3 verticle dots > Saved Searches in the top-right corner. You cannot delete a saved search created by another user.

Export AutoDiscovery Data to CSV

Click Options in the top right corner.
Click Export .