Cuesense by Viroic can help you finding new sales opportunities, identifying customer advocates, or simply creating structure in unstructured text sources. Interested in testing our text analysis technology? We offer several options for evaluating our sentiment analytics and entity extraction.

Purchase Intent analysis of Travel & Accommodations tweets

Cuesense Sentiment Analysis technology helps identifying new sales opportunities in corporate databases and social media. We applied this technology to the Travel & Accommodations domain and showed that we can deliver 165,000 purchase intent expressions from Twitter each day. These expressions could be used for building a consumer preference profile or initiating a sales transaction.

Cuesense collected 2770 tweets on the topic of travel and accommodations and analyzed the data set using its Sentiment Analysis engine. We searched for 18 representative keywords like “lodging” and “meeting venue” and collected data for 26 seconds. A first-level filter was applied that eliminated spammy messages from bots.

Show Query Filter

#merge -twitter:status:Source:”foursquare” -twitter:status:Source:”” -twitter:status:Source:”twitterfeed” -twitter:status:Source:”” -twitter:status:Source:”Nxy
#merge -RT
apartment “long weekend”
banquet venue
“close to airport”
conference venue
“extended stay”
holiday apartment
meeting venue
“near airport”
vacation package
wedding venue

The sentiment analysis engine looked for tens of thousands of text patterns and dozens grammatical constructs that represent opinions and emotions. Experts reviewed each tweet and evaluated the correctness of the algorithmic classification as correct or incorrect, focusing on purchase intent.

2770 tweets collected over 26 seconds indicate a potential for 9.2 million T&A-related tweets per day. Seventy-six (76) tweets contained statuses indicating that the sender was looking (having “Purchase Intent”) for accommodations or travel. This represents a 2.7% yield of commercially useful tweets for Purchase Intent and a rate of 248,000 “PI” tweets per day. Additionally, the algorithm is capable of identifying tweets useful for Testimonials and finding Competitive Switchers (not focus of this analysis).
Cuesense was able to correctly identify 50 of the 76 tweets, representing a 66% positive identification rate and 34% false negative rate. Cuesense has also flagged 12 additional tweets as “PI”, indicating a 19% false positive rate or 81% completeness rate. In other words, the algorithm found two out of three tweets containing Purchase Intent and four out of five tweets the algorithm flagged as having “Purchase Intent” indeed had this quality.

Overall, we project that Cuesense could deliver 165,000 Purchase Intent expressions related to Travel & Accommodations per day. These expressions could be used for building a consumer preference profile or initiating a sales transaction.

Key statistics

Statistic Value
T&A tweets, per day (est.): 9,200,000
Purchase Intent tweet yield: 2.7%
Purchase Intent tweet volume, per day (est.): 248,000
Cuesense Purchase Intent positive ID rate: 66%
Cuesense Purchase Intent completeness rate: 81%
Cuesense Purchase Intent false negative rate: 34%
Cuesense Purchase Intent false positive rate: 19%
Cuesense Purchase Intent expressions, per day (est.): 165,000
False positives, per day (est.): 37,000

Indicators of commercial suitability
Purchase Intent: Text must contain Anticipation or Interest emotions.
Testimonial: Text must have a positive valence (emotional orientation).
Competitive Switcher: Text must contain Disgust or Anger emotions.

Download the analysis Excel workbook to review the results of the analysis. Double-click on the Correct or Incorrect count cells to see a drill-through report. Search for the Status_Text field to see the original tweet text. Review the field Status_AnnotatedText for an in-depth view of how Cuesense identified sentiments and entities.