AI sentiment analysis in Tribe helps you understand the emotional tone of email communication with your customers. It supports better customer interactions, saves time, and enables insight through views, widgets, and automations. While other guides explain what sentiment analysis is and how to use it, this article focuses on how it works behind the scenes. By following this guide, you will understand how sentiment is calculated, stored, and structured within Tribe.
Table of Contents
What Does AI Sentiment Analysis Do?
How Is Sentiment Calculated?
Aggregated Sentiment And Trends
Understanding The Sentiment Data Model
Points To Consider In The Sentiment Data Model
Quick Summary
What Does AI Sentiment Analysis Do?
AI sentiment analysis automatically analyses the tone of emails in Tribe. Each email receives a sentiment score that reflects how positive, neutral, or negative the message is. These scores are used to calculate overall sentiment for activities, relations, and contacts. This allows sentiment data to be used in views, widgets, and automations.
How Is Sentiment Calculated?
When an email is created in your Tribe environment and sentiment analysis is enabled, an automated process starts immediately.
Analyse the content of the email.
Create a Sentiment Measurement record.
Link the sentiment measurement to the Email, related Activity, Relation, Contact, and optionally Employee.
Each sentiment measurement contains a score that represents the tone of that specific email.
Aggregated Sentiment And Trends
Every time a new incoming email is analysed, Tribe recalculates an aggregated sentiment score.
This aggregated sentiment:
Represents the current overall sentiment of an activity or relation
Is recalculated after each new incoming email
Helps track sentiment trends over time
The aggregated sentiment score is calculated using the following formula:
Aggregated Sentiment Score = (Previous Aggregated Sentiment Score + New Sentiment Score) / 2
This means the most recent email has more influence than older messages.
Tribe also analyses outgoing emails. These sentiment scores are not shown directly in the interface, but they are available for use in views, widgets, and automations. This can help analyse automated emails or support employee communication training.
Understanding The Sentiment Data Model
Sentiment analysis data is stored in a structured and reusable way within Tribe.
An incoming email is analysed.
A linked Sentiment Measurement record is created.
The Score field is filled with a value between -1 and +1:
+1 = Very positive
+0.5 = Positive
0 = Neutral
-0.5 = Negative
-1 = Very negative
This Score field can be used directly in widgets or views.
Each email is linked to a parent activity, such as a Sales Opportunity or Support Ticket. Based on the email score, an aggregated incoming sentiment measurement is created for that activity. The combined score is translated into a sentiment value from the sentiment list and shown in the activity header.
The same process applies to:
Relations (organisations or private individuals)
Contacts linked to those relations
Each receives its own aggregated incoming sentiment value, displayed in the header.
Points To Consider In The Sentiment Data Model
Tribe uses a generic data model, and sentiment analysis follows the same principle. For example:
An email is also an activity
A contact is also a relation
This flexible structure allows Tribe to adapt to different business processes. However, it also means you may see entities in the field picker that are not always populated.
For example, when viewing an email record, you may see linked entities that are not filled. These entities are only populated on the parent activity, such as the sales opportunity the email is linked to. Because emails share the same activity structure, the entities are visible but not always applicable.
Quick Summary
AI sentiment analysis in Tribe analyses email tone and stores results as sentiment measurements. These measurements are aggregated to show overall sentiment for activities, relations, and contacts. Understanding the data model helps you use sentiment data effectively in views, widgets, and automations.