BUSINESS SITUATION
The client’s key objective was to gain a comprehensive understanding of the overall voice of customers for themselves and their competitor’s OTT content through Sentiment Analysis.
SGA APPROACH
We have used three different approaches:
- Calculated the net sentiment score. We used the pre-trained model ‘roBERTa’, which is proven to be quite accurate for any type of textual comment
Tweet – “I loved the appearance, but the language was disgusting”
Output –
Negative 0.63942665
Neutral 0.24854486
Positive 0.112028554
- Used unigram-to-trigram mapping as a topic to subtopic mapping to understand the adjective of the conversation
- Generated a word cloud representing the most frequently-used words as the voice of customers, which allowed us to visualize the distribution of the topmost frequently-used expressions among customers
By utilizing the Fan Sentiment Model, we gave our client a meaningful analysis of competing shows
ENGAGEMENT
We utilized this model to produce a weekly report on streaming app buzz insights. This report highlighted the perceptions and feedback of customers and provided recommendations on how to act on their suggestions through an enhancement plan
BENEFITS & OUTCOME
The Fan Sentiment Model helped us get an understanding of the overall sentiment toward OTT platforms by tracking changes in brand reputation over time, which helped in identifying the areas that needed to be improved
KEY TAKEAWAY
- Sentiment Analysis helped the client to identify the content that resonates well with their audience and avoid the ones that could harm their brand reputation
- The distribution of the most frequently-used terms and their mapping helped stakeholders to easily understand the context revolving around the platform
- It also helped in uncovering important patterns and trends, enabling stakeholders to make more informed decisions, and taking targeted actions based on the insights gained.
