In any organization, email is considered as an exclusive mode of communication with the customers. On a regular day, when you receive countless emails it would be helpful to know the tone of the emails, because usually the ones that have a negative tone are given immediate attention over others.
Zia can help you prioritize the emails that have a negative tone by identifying the sentiment of the incoming emails. You can either choose Zia's pre-defined sentiment prediction model or build your own data model to train Zia on determining the tone of an email.
Note
- This feature is available only in the Ultimate Edition.
- Users with Administrator profile can configure Email Sentiment.
Activate Email Sentiment Analysis
Before you begin email sentiment configuration, make sure that you are logged in to your CRM account with Administrative privileges and have activated Email Sentiment.
To activate email sentiment configuration
- Go to Setup > General > Zia > Email Sentiment.
- In the Email Sentiment page, toggle on the Email Sentiment button.
Choose Existing Model
In the current pre-defined data model, Zia refers to a spectrum of emails from the IT sector and determines the tone of incoming emails. You can evaluate Zia's sentiment prediction to see if it suits your company's requirements, by adding a few test emails from your organization.
Importing test emails - You can import a few sample emails received form your customers or clients and check if Zia's sentiment prediction is accurate. Based on the number of emails that are predicted accurately by Zia, an accuracy measure is generated. In case, you are unsatisfied with the accuracy measure you can choose to create your own sentiment prediction model.
To choose Zia's existing model and add test emails
- Go to Setup > General > Zia > Email Sentiment.
- In the Email Sentiment page, click Zia Existing Model.
- Click Import Testing Data.
- Choose a file and click Import.
- Click Publish Now, to start using the existing sentiment prediction model.
Create Your Data Model
When Zia's pre-defined sentiment prediction model does not comply with your organization's requirements, you can create your own model. You can select a few emails received from the customers or clients as a sample to train Zia in predicting the sentiment of the emails.
- Adding training data - You can import a set of sample emails and based on your assessment, can mark them either positive, negative or neutral. Zia will use these emails as sample and learn to predict the sentiment of the emails.
- Adding testing data - After you have added the training data, you can evaluate Zia's prediction by using some sample test emails. You can import a list of test emails and gauge Zia's sentiment prediction. In case, you are unsatisfied with Zia's results you can add more training data to help Zia predict better.
To add training data in your organization's data model
- In the Email Sentiment page, click Your Own Data Model.
- In Training Data, click Import Training Data.
- Choose a file and click Import.
To add testing data
- In the Email Sentiment page, click Your Own Data Model.
- In Testing Data, click Import Testing Data.
- Choose a file and click Import.
- Click Publish Now, if you want Zia to use this data model for sentiment prediction.
Add Sample Emails
Instead of importing the sample test emails, you can type the email content and choose a sentiment based on your assessment and add it as a record.
To add sample emails as records
- In the Email Sentiment page, click Add record.
- Type the email content.
- Select a sentiment from the drop-down list.
- Click Done.
Note
- You can add up to 10,000 testing data or training data into your CRM account.
- You can import a maximum of 5000 records and the maximum file size should not exceed 5MB.
- An import file must be in csv, xls or xlsx format only.