A Global Oral Hygiene Brand Performs Sentiment Analysis on Consumer Reviews with MediaAgility’s NLP Based System
Headquartered in the United States, the Customer’s line of oral hygiene products is widely popular across the globe. To be competitive in their global markets, the Customer wanted to read into consumer reviews, assess their consumers’ sentiments, and base their business strategies on reliable and relevant insights.
Transforming the Brand With Consumer Insights
Earlier, the Customer manually ran scripts to download consumer reviews from various social media sites. Then, they performed sentiment analysis on the downloaded data. However, they could not visualize the results of the analysis on a centralized dashboard and take effective and timely well-informed decisions.
With MediaAgility, their digital consulting partner, the Customer built a sentiment analysis system to process and derive insights from consumer reviews. MediaAgility developed a custom solution to enroll data from two data streams – consumer reviews and Twitter tweets. The data stored in BigQuery is then ingested into the system for analysis on DataFlow jobs.
With NLP, reviews and tweets are processed and assigned an overall-brand-sentiment score and a sentiment-by-product score. DataStudio dashboards facilitate easy access and analysis of the NLP results.
Google Cloud products facilitate the following functions of the system –
- Cloud Composer: Cloud Composer orchestrates the workflow and schedules the DataFlow jobs
- Google Cloud Pub/Sub: Pub/Sub ingests the raw tweets streamed in by Twitter streaming APIs
- Google Kubernetes Engine: GKE captures tweets at scale and streaming inserts them into BigQuery
- Google Cloud Storage: Reviews from the DataFlow scrapping jobs are dumped here
- Google Cloud DataFlow: DataFlow has multiple batch pipelines to perform transformation/scrapping of social media data
- Google Cloud Bigquery: Raw reviews and tweets, NLP, and sentiment data are stored here
- Google Cloud DataStudio: Reports are generated from raw data and sentiment-analysis data stored in BigQuery
With sentiment analysis and the dashboard, the Customer has been able to –
- Run scripts automatically with NLP
- Improve on their products based on consumer feedback
- Assess the overall brand sentiment
Currently, based on the insights from the analysis, the Customer is reevaluating their brand positioning and also planning various campaigns to increase the reach and awareness of their products.