Yesterday, February 26th, was a day of firsts for us at NetElixir University! Our CEO, Udayan Bose, delivered his first webinar focused on social media, when he presented an original NetElixir study analyzing the feelings and emotions of consumers by looking at their tweets. In this webinar, he demonstrated how we performed our Social Sentiment Analysis, and how businesses can do the same to figure out the sentiments behind their own consumer’s tweets about their business.
In this original study, we gathered tweets from February 9th to the 15th by tracking Valentine’s Day related hashtags (such as #ValentinesDay) at a rate of approximately 425 tweets an hour! We then looked for things such as promo text and retweet patterns to identify tweets that were posted by marketers, and not consumers, which ended up being about 65% of the tweets collected. Shockingly, we found that out of the Valentine’s Day tweets posted by marketers, about 62% of these tweets had no engagement at all. It makes us wonder—is Twitter becoming more of a Sales platform, compared to an engagement one? This is definitely something we’ll be keeping an eye on in the future.
After pulling out the marketer’s tweets, we narrowed our tweets down to a sample of about 25,000 consumer tweets centered on Valentine’s Day. Here’s where we got into the nitty-gritty to find out the feelings of these consumers. First, we started by importing our collected data into a statistical computing tool called R. Next, we converted words into their root form (such as “loved” to “love”) to make the data easier to understand. After that, we changed all of our collected tweets to lowercase, and stripped out their punctuation, usernames, and links. Lastly, we removed “stop” words (such as “the”), as they didn’t contribute to our analysis. Finally, we were ready to dig into this data!
Here is where our Social Sentiment Analysis, which we uploaded into R, came into play. Our sentiment analysis is a way of figuring out if a word had a positive or negative connotation (such as “love” would be positive, and “hate” would be negative). We also uploaded a word frequency algorithm into R, to see what products, events, and ideas people were talking about the most.
So, what did we find? We found that common topics people talked about doing for Valentine’s Day were going out to a V-Day dinner, going to Disneyland, getting some Starbucks coffee, and getting their nails done before a date! We also found many couples talked about listening to Spotify together. As for television and movies, the big blockbuster winner was 50 Shades of Grey, while many people were also talking about the “Galentine’s Day” episode of the comedy Parks and Recreation, which aired its series finale this Tuesday.
As for the most talked about products, the winner by a landslide was chocolates, with flowers and greeting cards close behind. Other commonly talked about products were rose petals, cake, jewelry, and diamond rings. We also found that although most tweets about chocolate were being talked about in a neutral tone, the positive tweets about chocolates were very positive (which doesn’t surprise us— yum!)
While our sample was taken from Valentine’s Day consumers, our Social Sentiment model can easily be applied to your specific business’s consumers to gauge how connected your consumers are with your brand. We believe this model is a great way to gain some insight into what your top advocates (and critics) are saying about your brand, although Twitter alone may not be enough to paint the full picture. Looking into other social channels is something we plan to experiment with in the future, to gain more insight to what consumers are really saying.
To watch this webinar again, or any of our other webinars, check out our NetElixir University page!
You can also watch this webinar on our YouTube channel, or check out just the slides on Slideshare. We had a blast delving into the analytics behind Social Media, and look forward to exploring it more and sharing our insights with you!