Marc's Augmented Reality

It is extremely important for companies to participate in the conversation taking place on blogs, forums, Twitter, Facebook, etc. to protect their brand.  The problem is there are so many conversations taking place it can quickly become overwhelming.  This is where sentiment analysis can help.

Sentiment analysis is the ability to automatically collect and categorize what is being said about you based on customer sentiment.  Then you can focus your attention on mitigating the negativity directed at your product or services.

There are quite a few sophisticated sentiment analysis solutions like Scout Labs and Jordange.  To give you a feel for this technology I will focus on the ligher weight solutions that in this case just monitor sentiment on Twitter.

Tweetfeel

The first tool we will look at is Tweetfeel.  I entered “Starbucks” in the search field and here are the results.

Tweetfeel analyzes each tweet meeting the search criteria highlighting the sentiment in the tweet using red and green.  As you would guess, green is positive and red is negative.  The total count of each type is displayed below the smile and frown faces above the tweets.

Twendz

To get started with Twendz I enter “Starbucks” in the search box at the upper right.  It immediately began to processing the tweets based on the speed you select in the slider at the top (default is Medium).  Being able to slow down the feed provides you time to read the tweets as they are processed.

The overall sentiment is shown in the group box labeled TOPIC again using the red for negative and green for positive sentiment.  The white represents neutral sentiment.  In addition, it displays SUBTOPICS and a word cloud.  Twendz has a nice interface.

Twitrratr

Every time I read this name I have to laugh thinking of how Arnold would say it with his accent!  Once again I enter “Starbucks” in the search box and click SEARCH.

Twitrratr shows the results in three columns; positive, neutral, and negative.  Totals are shown at the top.

Twitrratr appears to be the least sophisticated of these solutions associating sentiment based on a single word.  ”Tired” must be a negative word as in “I’m so tired of paying so much at Starbucks.”  But in the case of this tweet “i’m hot and tired from rehearsal!” the sentence is neutral.

Conclusion

Doing a quick calculation based on current tweets, “Starbucks” is mentioned close to 8,000 times per day.  It would be very time consuming to review each of these tweets to determine how to respond.  With sentiment analysis tools you can focus your attention on the 300 or so negative tweets … much more manageable.