Configuring Conversation Analyzer in Salesforce

If Advanced Reports and Statistics is enabled and configured for your account, you can use Conversation Analyzer in Salesforce.

Conversation Analyzer adds the following features to your Salesforce org:

• Voice visualization
• Event timeline
• Transcript of the recording
• Categorization of the contents of the recording
• Transcript confidence score (High/Medium/Low)
• Sentiment summary and scores, if sentiment analysis is enabled for your account

If enabled for your account, to use Conversation Analyzer in Salesforce, you must perform various configuration tasks.

Prerequisites for configuring Conversation Analyzer in Salesforce

Before you can configure Conversation Analyzer in Salesforce, you must install and configure Advanced Reports and Statistics. For information about Advanced Reports and Statistics, see Advanced Reporting and Statistics in Salesforce.

How do I make Conversation Analyzer available in Salesforce?

To see Conversation Analyzer components in NVM Call Summary records, change the page layout that the NVM Call Summary object uses:

• For inbound calls, use the NVM Analytics Summary layout
• For outbound calls, use the NVM Analytics Outbound Summary layout
• For internal calls, use the NVM Analytics Internal Summary layout

The top section of the page layouts embeds Vonage Contact Center's Conversation Analyzer into an NVM Call Summary record. For information about the contents of Conversation Analyzer, see Conversation Analyzer.

Other sections on the page layouts are customized for the call type, containing only the fields that are relevant to those calls.

When you or an agent views a NVM Call Summary record, Conversation Analyzer for the related call appears. You can see Conversation Analyzer in the Conversation Analyzer section of the record.

How do I retrieve call recording analytics data or transcripts, or both, using Advanced Reporting and Statistics?

As well as retrieving call summary and agent state statistics, Advanced Reporting and Statistics can retrieve call recording analytics data from Vonage Contact Center. Advanced Reporting and Statistics can also retrieve call recording transcripts. Advanced Reporting and Statistics stores the additional data in fields on the NVM Call Summary object.

To retrieve call recording analytics data or transcripts, or both, perform the following steps:

1. Go to Setup in Salesforce.
2. On the left-hand side of the Salesforce window, in the Quick Find field, type 'Installed Packages'.
3. In the filtered list of items, click Installed Packages. Installed Packages appears.

4. Locate and click NVM Advanced Reporting. NVM Advanced Reporting (Managed) appears.

5. Alongside Post Install Instructions, click View.

6. In the Global Settings and Mappings section, in Flags, type 'SPEECH'.

   Optional flags

   You can add optional flags CAROWS and CA(0). For information about these flags, see Configuring when Vonage Call Summary records are created and Configuring Conversation Analyzer Insights records later in this page.

   Click Save All. Fetch Analytics and Fetch Transcript check boxes appear in the Channels section.

7. Alongside each channel, select check boxes as required:

   • Fetch Analytics. Select the Fetch Analytics check box to retrieve and store the following data:

     Data to retrieve	Fields to store data in	Notes
     Category and subcategory data	Custom category and subcategory fields. Because you can customize categories and subcategories, you must create the fields yourself. For information, see Creating custom fields for categories and subcategories later in this page. Matched categories also appear in the Categories (NVMStatsSF__Categories__c) field.	If you do not provide all related category and subcategory fields, Advanced Reporting and Statistics reports an error and does not retrieve applicable category and subcategory data.
     Percentage of agent talk time	Agent Percent (NVMStatsSF__Agent_Percent__c)	Fields appear in the Caller and Agent Related section. If the field do not appear, add them to the relevant NVM Call Summary page layouts.
     Percentage of talkover time	Crosstalk Percent (NVMStatsSF__Crosstalk_Percent__c)
     Percentage of customer talk time	Customer Percent (NVMStatsSF__Customer_Percent__c)
     Percentage of silence time	Silence Percent (NVMStatsSF__Silence__c)
     Level of confidence	Confidence (NVMStatsSFConfidence__c)
     Language of call	Language (NVMStatsSFlanguage__c)
     Sentiment summary and scores	Sentiment (NVMStatsSF__Sentiment__c)	Field appears in the Information section in the Vonage Call Summary object if sentiment analysis is enabled for your account. If the sentiment field does not appear, add it to the relevant NVM Call Summary page layouts.

   • Fetch Transcript. Select the Fetch Transcript check box to retrieve call recording transcripts. The transcript is stored in the Call Transcription field and is searchable within Salesforce.
8. Click Save All. Next time a transaction runs, it uses the new settings and retrieves additional analytics data. Advanced Reporting and Statistics populates NVM Call Summary record fields as appropriate.

Configuring when Vonage Call Summary records are created

From Advanced Reporting and Statistics version 2.84 onwards, you can add the CA(0) flag. This flag determines how quickly the package should get data for Vonage Call Summary records. By default, the package waits 3600 seconds between Vonage Call Transcription, Categorization, and pulling the resultant data into Salesforce. CA(0) instructs the package to override the default wait of 3600 seconds with 0 seconds. Because Conversation Analyzer data is checked every 15 minutes, this will typically load the analytics data into your org 1–16 minutes after the data analysis is completed. You may use other values but 0 is recommended.

Configuring Conversation Analyzer Insights records

From Advanced Reporting and Statistics version 2.84 onwards, you can add the CAROWS flag. This flag instructs the package to create child Conversation Analyzer Insights records under a Vonage Call Summary record. A Conversation Analyzer Insights record is created for each combination of category or subcategory matches for each call. This avoids the need to create custom fields for categories and subcategories.

The maximum number of rows created for each call — if all your defined Conversation Analyzer categories are matched at least once — is number of categories x number of subcategories.

Creating custom fields for categories and subcategories

If you choose to retrieve call recording analytics data from Vonage Contact Center, you can create custom category and subcategory fields on the Vonage Call Summary in which to store that data. (The Advanced Reporting and Statistics managed package provides custom fields to store other analytics data.) Each category and subcategory can have a corresponding field.

We recommend that you create a custom checkbox field for each category and a number field for each subcategory that you have configured for the Conversation Analyzer feature. Set Field Name—as referenced by APIs—for the fields using the following format:

• Categories: Category__c
• Subcategories: Category_subcategory__c

For example, if you have a Communication category and an Active listening subcategory, create fields with the following types, labels, names, and default values:

Optionally, create a section on the summary page layouts and add the category and subcategory custom fields.

Next time a transaction runs, Advanced Reporting and Statistics retrieves category and subcategory data and populates custom fields as appropriate. A checked field indicates that the transcript of the call recording matched the category or subcategory.

Creating custom fields for sentiment percentages and scores

If sentiment analysis is enabled for your account, you can configure Advanced Reporting and Statistics to retrieve sentiment summaries and scores along with call recording analytics data or transcripts, or both. For information about retrieving sentiment data, see the Fetch Analytics table earlier in this page. If you retrieve this data, the Sentiment field will contain a string of concatenated sentiment data. For information about this data, see Analyzing a call recording in Conversation Analyzer in Salesforce.

To make this data easier to read, or to report on it, you can separate the data into multiple fields. The following code samples extract data from a Sentiment field, for example:

Overall score

To extract the overall score from Sentiment and put it into a separate custom field, you must create a formula-type custom field that uses the provided formula. Set the Field Name for this custom fields to OverallSentimentScore__c and Field Label to Overall Sentiment Score. Use the following formula. The formula returns a number value.

VALUE(
  TRIM(
    LEFT(
      SUBSTITUTE(
        SUBSTITUTE(NVMStatsSF__Sentiment__c, "Overall:", ""),
      ",",
      " "),
    6)
  )
)

Sentiment percentages

To put percentages of incidences of very negative, negative, neutral, and so on, sentiments into individual fields, firstly, create custom fields. You'll need a formula-type custom field for each of the percentages you want to save. Set Field Names for these custom fields to VeryNegativeSentiment__c, NegativeSentiment__c, NeutralSentiment__c, and so on, and Field Labels to Very Negative Sentiment, Negative Sentiment, Neutral Sentiment, and so on.

The following formulas return number values.

Use the following formula for the VeryNegativeSentiment__c field:

VALUE(
  TRIM(
    LEFT (
      SUBSTITUTE (
        MID(NVMStatsSF__Sentiment__c,
        FIND("VeryNegative:", NVMStatsSF__Sentiment__c)+13,
        6),
      ",", "      "),
    6)
  )
)

Use the following formula for the NegativeSentiment__c field:

VALUE(
  TRIM(
    LEFT(
      SUBSTITUTE(
        MID(NVMStatsSF__Sentiment__c,
        FIND(", Negative:", NVMStatsSF__Sentiment__c)+11,
        6),
      ",", "      "),
    6)
  )
)

Use the following formula in the NeutralSentiment__c field:

VALUE(
  TRIM(
    LEFT (
      SUBSTITUTE (
        MID(NVMStatsSF__Sentiment__c,
        FIND("Neutral:", NVMStatsSF__Sentiment__c)+8,
        6),
      ",", "      "),
    6)
  )
)

Use the following formula in the PositiveSentiment__c field:

VALUE(
  TRIM(
    LEFT(
      SUBSTITUTE(
        MID(NVMStatsSF__Sentiment__c,
        FIND(", Positive:", NVMStatsSF__Sentiment__c)+11,
        6),
      ",", "      "),
    6)
  )
)

Use the following formula for the VeryPositiveSentiment__c field:

VALUE(
  TRIM(
    LEFT (
      SUBSTITUTE (
        MID(NVMStatsSF__Sentiment__c,
        FIND("VeryPositive:", NVMStatsSF__Sentiment__c)+13,
        6),
      ",", "      "),
    6)
  )
)

Sentiment scores

To extract each of the individual scores from Sentiment and put them into separate custom fields, you must create ten formula-type custom fields. Set Field Names for these custom fields to SentimentScore0__c, SentimentScore1__c, SentimentScore2__c, and so on, and Field Labels to Sentiment Score 0, Sentiment Score 1, and Sentiment Score 2. Use the following formula in the individual sentiment fields, replacing n with the specific sentiment score position. The formula returns a number value.

VALUE(
	MID(RIGHT(NVMStatsSF__Sentiment__c,59) , 1+6*n, 5)
)

For Sentiment Score 0 field, use the following formula:

VALUE(
	MID(RIGHT(NVMStatsSF__Sentiment__c,59) , 1+6*0, 5)
)

For Sentiment Score 9 field, use the following formula:

VALUE(
	MID(RIGHT(NVMStatsSF__Sentiment__c,59) , 1+6*9, 5)
)

Individual sentiment examples
This example uses the following value in the Sentiment field:

Using the formula above in the Sentiment Score 0 field would result in the following value:

Using the formula above in the Sentiment Score 9 field would result in the following value: