Reliable Sentiment Score

August 10, 2018 / by Shiho Hashimoto Shiho Hashimoto


In my previous blog, I explained how you grade on sentiment score of your brand on social media can either help increase sales or bring unforeseen crisis. But how reliable can it be?

To make sure your sentiment score is reliable, you need to evaluate the quality of your current analytics technology.  Watch this 2-minute video to learn at least 5 things you just can’t ignore. Transcript of the video is available below, in case you prefer reading over watching/listening!


Screen Shot 2018-08-10 at 10.57.35


Your brand's social sentiment score must be accurate because it helps you formulate your marketing strategy to increase sales revenue. To make sure your sentiment score is reliable, you need to evaluate the quality of your current analytics technology. There are at least 5 things you just can’t ignore.

1. Removing irrelevant data

Sentiment analysis process starts with collecting data. From social media platforms, you will get huge amount of unfiltered data. Make sure all irrelevant data created by bots, internet trolls, and spammers is removed. Otherwise, your sentiment score will be incorrect.

2. Analyzing unstructured data

Data from social media is not always text-based or numerical. It also includes unstructured data, like videos, images, or even voice. Without analyzing these elements, your sentiment score will be skewed.

3. Understanding ambiguity

Many posts on social media contain both texts and visual elements, creating multiple layers of meaning. Sarcastic and ironic posts can only be interpreted correctly at the moment by human analysts; so relying only on computer algorithms is a mistake.

4. Detecting cultural norms, spelling mistakes, and trending slang words

Social media reflects our hyperconnected and ever-growing culture, where phenomena, newly-invented slang words and texting becomes trendy every day. These elements can set contexts for the mentions related to your brand and bring details about your audience. Your analytics tool needs artificial as well as human intelligence to keep up with these constant changes. Otherwise, you’re bound to lose accurate sentiment insights.

5. Understanding multi-language

When you are working with a global brand, you must make sure that your analytics technology can interpret sentiment data in different languages. Or else, you cannot understand all of your customers, perform customer service well enough, or secure your position in international markets.

At InsightsAtlas, we offer a service which combines the advanced social media data science with human intelligence to tackle these problems, providing our clients with accurate sentiment insights and reinforcing their marketing strategy. Thanks for watching this video and don’t forget to subscribe to our channel!

Contact Us!

Tags: Brands, Marketing, Social Analytics, Social Media, Partner Benchmarketing, Social Media Consultation, Brand Ambassadors, Brand Insight, Brand Management, Communication, Crisis Management, Big Data, Influencer Analysis, Influencer Marketing, Sentiment Analysis, Social Listening, Brand Visibility Metrics, Trend Tracking, Organic Conversion Metrics, ROI Measurement, Competitor Benchmarking, Audience Response, Social Media Alerts, Landscape Tracking, Social Media Monitoring, Brand Health Index

Share This!

Recent Posts

Subscribe to Our Blog!