Social Media Listening

Brief

An established baby product manufacturer whose baby products were sold through retailer wanted a study on the customer sentiment

Challenge

The client’s product team had no clear clarity on the popular themes in the targeted market related to its baby products. The products were designed or added in the existing portfolio on the basis of judgement of the core design team – which was not backed by deep data insight.

This core design team/ management randomlyused to go throughreviews on the company’s baby productsleft on social media, forums, and e commerce sites – but were not able to discern something concrete based on the same.

Hence, they wanted us to do the studyto corroborate if their management/ design team thinking was in line with theprevailing taste and preferencein the market.

Our Action

• We collated the Tweets and other posts on the social media through API
• Framed the untrained rules along with the  random variables
• As we got the robust data, we applied  Naive Bayes classification algorithm
• The gathered data was pre-processed to lower the load on the memory for example – stemming, stop wordsremoval. ; we also checked for bias/ unbalanced data
• Multiple Bayesian Models were developed for getting answers to different queries that our team preparedin response to client expectation out of this study
• We finally developed the dashboard for client to take necessary decisions based on the study

Results

• Product team planned action plan to reduced negative sentiments discovered in the study
• Client could  take superior  decisions on new product launch based on gathered data insights on customers