Sentiment Analysis Rocks! Here’s Why.

Subscribe to our monthly newsletter to get the latest updates in your inbox

Ah, sentiment analysis. Now is the time where I typically use a terrible Dad pun and talk about how “sediment” analysis is determining how much clay and other elements are in a sample. Of “quartz,” given the eye roll that I know my kids would give me, I’ll skip the geology humor. At its core (I just can't stop), sentiment analysis is the interpretation and classification of emotions.

How to Get Started

With Google Cloud Natural Language Processing (NLP), sentiment analysis is now easy, scalable and low cost. Previously, putting scientific rigor behind a sentiment analysis exercise took a great deal of time and required several resources to classify all of the customer comments about a business’s services, products or brand. Now, with the help of Google Cloud, it is possible to quickly bring this scientific rigor to millions of client statements in order to help monitor and determine client sentiment.  Here’s a sample Data Studio report that we created to show public sentiment. For this illustrative exercise, we used the HackerNews public BigQuery dataset in conjunction with Natural Language API to see how the audience’s comments reflected the sentiment of various public clouds.  For those of you not familiar with Hackernews, it is a website focused on computer science where users post stories and content. Toward the bottom of the Data Studio report, we created a small section where you can see the cost to run this report and how many stories it is analyzing. For a cost of under $10 a month, we are analyzing and storing data on almost 400,000 rows of content.  Hopefully, this helps solidify (I have a problem) the case that this solution is very scalable. The obvious next questions are: what can you do with sentiment analysis, and why is it useful? 

Why You Should Use Sentiment Analysis

Imagine a company that announces a new product, and customers start to comment (tweeting, blogging, etc) about it. A marketer can now evaluate these comments and see if new features need to be emphasized or repositioned based upon the customers’ sentiments about the new release.   Let’s go back to our Data Studio example and filter the public clouds to only include GCP content. You can imagine Google looking through this list to see how new features and products are trending.  For example, this article is automatically highlighted as it has 11 comments that were measured of mostly negative sentiment. Looking at the article closer, it tells a really positive story about how to use a Google Cloud Platform product called Cloud Armor.  However, many of the comments talk about how hard it is to install and use. You can see how this sort of insight could be valuable to Google from both a content and a product perspective. Each insight influences future messaging and helps educate the product team on how features are being perceived. 

Additional Use Cases

As digital interactions grow, the use cases are expanding exponentially. Below is a list of different sentiment analysis projects we have helped customers with:
  • Amazon/Product reviews
  • Correlating support chat to website behavior
  • Competitive products
  • Brand monitoring
The use cases are almost infinite, and given the truly scalable nature of NLP on the Google Cloud Platform, all businesses should be thinking about how to use this technology. We encourage you to interact with our sample Data Studio report — also shown below — and contact us to see how we can help make sentiment analysis a bedrock of your business. (Okay, I’m really done now.)