Analyzing Customer Sentiment and Business Opportunities by Detecting Emojis and Understanding Text.

GOAL

  • Daily analysis of all internal and social media sources for customer sentiment 
  •  Categorize them in five levels of satisfaction/dissatisfaction 
  • Analyze customer satisfaction at various hierarchical levels and by age, profession, gender, and geography.  
  • Actions for each level of satisfaction. 

CHALLENGE

For this young, fast-growing, living spaces company customer satisfaction and wordofmouth promotion was extremely important for growth. They wanted to know their customers’ satisfaction and reaching out to their network 

SOLUTION

WE CAME, WE SAW, WE SOLVED

We used three internal sources and 8 external social media sources to consolidate data at various hierarchical levels and categories. Then our data science team created a machine learning model to understand emotions in textual data, converting texts and emojiinto tokens, identifying emotions from those tokens. The team also created an action plan, action templates, and reports based on the findings. 

The model continues to finetune as we enter second year of implementation.

RESULTS

  • Referral business grew by almost 3x.
  • Customer turnover reduced by 22%.
  • Management could target their marketing budget for higher customer success.
katapultSentiment Analysis