Reclassifying Listening Sentiment

This feature is only available for Social Listening customers.

This is a guide on how to improve sentiment accuracy in Sprout’s Social Listening by reclassifying message sentiment either manually or in bulk by using Sentiment Reclassification and Sentiment Rules.

What is Sentiment Reclassification?

Sentiment Reclassification is a feature in Sprout Social's Listening tool that enables you to manually correct the sentiment of messages. While Sprout's AI-powered sentiment analysis is a powerful tool for gauging audience perception, the nuances of human language mean that it can sometimes misinterpret the intended sentiment of a message. With Sentiment Reclassification, you can apply your brand's unique context to refine the sentiment data, leading to more accurate insights.

When Should I Use Sentiment Reclassification?

You should use Sentiment Reclassification when you notice that Sprout's Listening tool is consistently misinterpreting the sentiment of certain messages. This is particularly useful for brand names, product names, or industry-specific jargon that may have a positive, negative, or neutral connotation that is unique to your business.

For example: If you're a gaming company with a new release called "War Zone," the Listening tool might incorrectly flag mentions of this title as negative because the term “war zone” is inherently negative. To correct this, you can create a Sentiment Ruleset to treat "War Zone" as a neutral phrase. This prevents your analytics from being skewed by false negatives and gives you a more accurate picture of your brand's health.

How to Reclassify Sentiment for a Single Message?

If you come across a single message with an incorrect sentiment classification, you can change it by following these steps:

1. Navigate to your Listening Topic.
2. Open the Messages tab and click on the message you want to reclassify.
3. Click the dropdown menu in the message details view.
4. Reclassify Sentiment by choosing the correct sentiment (Positive, Neutral, or Negative).
    

How to Reclassify Sentiment in Bulk?

For situations where multiple messages need their sentiment reclassified, you can use the bulk reclassification feature:

1. Select the messages within your Listening Topic you want to reclassify by clicking the checkboxes next to them.
2. Click the Reclassify Sentiment icon (a smiley face with a pencil) in the menu at the top of the message list.
3. Select the correct sentiment from the dropdown menu.

It may take up to a minute for message sentiment changes to take  effect.

Using the Message Explorer for Sentiment Reclassification

The Message Explorer provides a comprehensive view of all messages in your Listening Topic, making it an ideal place for sentiment reclassification. You can use the filters to narrow down your message list and then reclassify sentiments individually or in bulk, following the same steps outlined above.

To reclassify messages from the Message Explorer:

1. Select the messages you want to reclassify from the Message Explorer window.
2. Click the Reclasify Sentiment icon (a smiley face with a pencil).
3. Select Positive, Neutral, Negative or Unclassified.

Why are many of my messages showing as unclassified?

Messages may appear as "unclassified" for several reasons:

• Ambiguous Language: The message may contain language that is too ambiguous for the AI to confidently assign a sentiment.
• Mixed Sentiments: The message might contain both positive and negative elements, making it difficult to classify as one or the other.
• Unsupported Language: The message may be in a language that Sprout's sentiment analysis doesn't yet fully support.
• Lack of Context: The AI may lack the necessary context to understand the sentiment of certain brand names, product names, or industry-specific terms.

You can address this by using Sentiment Rules to provide the AI with more context, or by manually reclassifying the unclassified messages.

FAQs

What is sentiment reclassification in Sprout Social?

Sentiment reclassification is a feature in Sprout's Listening tool that allows you to manually correct a message's assigned sentiment (Positive, Neutral, or Negative). This helps ensure your data accurately reflects audience perception.

How do I fix an incorrect sentiment tag on a message?

To fix a single message, click on it within your Listening Topic, select the smiley face with a pencil icon in the corner, and then choose to Reclassify Sentiment from the dropdown menu options.

Can I change the sentiment for many messages at once?

Yes, you can use the bulk reclassification feature. From your message feed, select the checkboxes next to all the messages you want to change. A menu will appear at the top; click the Reclassify Sentiment icon and apply the correct sentiment.

Why are so many of my messages marked as "unclassified"?

Messages are often marked "unclassified" if the language is ambiguous, contains a mix of positive and negative phrases, is in a language not fully supported for sentiment, or if the AI lacks the specific context for a term (like a new product name).

What's the difference between reclassifying sentiment and creating a Sentiment Rule?

Reclassifying is a reactive fix for existing messages. Creating a Sentiment Rule is a proactive measure that tells the Listening tool how to classify all future messages that contain a specific keyword or phrase.

Can I reclassify sentiment for all my messages at once?

Yes. You can reclassify all messages (up to 1000) by clicking the checkbox at the top of the Message Explorer.

How long does it take for my reports to update after I reclassify sentiment?

Your Listening Topic data and reports will update almost instantly. The change in sentiment distribution will be reflected in your analytics as soon as you save the reclassification.

Can I undo a sentiment reclassification if I make a mistake?

Yes. You can follow the same reclassification process on the message to change it back to the original sentiment or select a new one.

Why was a positive comment about my brand marked as negative?

This can happen if the comment contains words that are typically negative, even if the intent was positive (e.g. "That new feature is wicked!"). Manual reclassification is the best way to correct this.

Is it possible to make my brand name always appear as neutral?

Yes, this is a perfect use case for Sentiment Rules. You can create a rule that defines your brand name as a neutral term, which will prevent it from being skewed by positive or negative adjectives around it.

Is there a log or history of all sentiment reclassifications?

The Sprout platform doesn't currently feature a specific audit log showing a historical list of manual sentiment changes. The change is reflected on the message and in the overall analytics.

Why don't I see the option to reclassify sentiment in my Listening Topic?

This is almost always a permissions issue. You need the "Manage" permission for Listening. Contact your team's Sprout administrator to have your user role updated.

How can I find all messages that need sentiment classification?

In the Message Explorer, use the filters on the left-hand menu to filter your view. You can select "Unclassified" under the Sentiment filter to see only those messages and address them in bulk.

Does changing sentiment in one Listening Topic carry over to another?

No. Both manual Sentiment Reclassification and Sentiment Rules are specific to the individual Listening Topic you are working in. They do not transfer to other Topics.

What are the best practices for maintaining accurate sentiment data?

Periodically review your messages, especially those marked "Unclassified." Use bulk reclassification for quick corrections and create Sentiment Rules for recurring phrases, brand names, or jargon to improve accuracy over the long term.

Want to learn more? Check out these articles:

• How does Sprout determine sentiment?
• What types of insights can I get from Social Listening?
• Social Listening FAQs