Sentiment Analysis uses natural language processing to identify important keywords in your customers' reviews and assign a sentiment score based on the context in which those keywords are used. This helps you understand which aspects of your business are driving positive and negative feedback without manually reading through reviews.

Navigate to Reviews > Sentiment Analysis to access the dashboard.

How Sentiment Analysis Works

Sentiment Analysis identifies two elements in each review:

• Keywords: the subjects being discussed, typically nouns such as "service," "food," "wait time," or "staff"
• Modifiers: the words describing those keywords, typically adjectives and adverbs such as "friendly," "slow," "excellent," or "broken"

Together, keywords and modifiers determine the sentiment score for each topic. For example, "friendly staff" produces a positive score for the keyword "staff," while "broken ATM" produces a negative score for "ATM."

Keywords are identified automatically from your existing reviews. You cannot add or remove keywords manually — they appear in Sentiment Analysis only after being identified in at least one review. Not all reviews are eligible for Sentiment Analysis depending on the publisher, content, and language.

Navigate the Dashboard

The Sentiment Analysis dashboard has three main sections. Use the filter toolbar at the top to filter by entity, date, publisher, and other dimensions. Adding a time filter shows period-over-period deltas for each metric.

Sentiment by Collection

Collections appear at the top of the screen. A collection is a group of related keywords that you create to track a theme — for example, a "Food" collection containing keywords like "burger," "fries," and "menu." Collections show the aggregate mention count, average rating, and sentiment score for all keywords in the group.

Sentiment Word Cloud

The word cloud displays up to 100 of the most frequently mentioned keywords. The size of each word indicates how often it appears in reviews. The color indicates its sentiment score using the same scale as the rest of the dashboard. This gives you a quick visual overview of the most prominent themes in your reviews.

Sentiment by Keyword

The keyword table shows every keyword identified in your reviews, with the following data for each:

• Sentiment Score: a score from -100 to +100 indicating how positively or negatively the keyword is discussed. Most keywords fall between -10 and +10. Scores outside that range indicate strong positive or negative sentiment.
• Mentions: the number of times the keyword appears across your reviews
• Average Rating: the average star rating of reviews containing the keyword
• Top Modifiers: the words most frequently used to describe the keyword

Click on a modifier to preview recent reviews containing that keyword-modifier combination. Click the highlighted link to open the full filtered list of reviews in the Monitoring tab.

You can sort the keyword table by mention count, sentiment score, or average rating.

Create a Collection

Collections let you group related keywords to see aggregate sentiment for a topic area. For example, a restaurant might create a "Service" collection containing keywords like "staff," "waiter," and "manager."

To create a collection:

1. Navigate to Reviews > Sentiment Analysis.
2. Click Create Collection.
3. Enter a collection name.
4. Click + Add Keyword and select keywords from the list.
5. Click Save.

You can also start creating a collection from a specific keyword by clicking the pencil icon to the right of that keyword in the keyword table.