how to analyze buyers sentiments of amazon products using customer review insights

Actionable perceptions from Amazon reviews to make better decisions

In the past two decades, the internet has got a revolution in shopping experiences. Having nearly everything accessible online, the customers can just open the app, do some tapping and that’s it, the product will be delivered at the doorsteps!

Though, with hundreds of available options in different categories, this can be an intimidating task to search for every option as well as choose the finest one. Mostly, product reviews are differential when the time comes to make an important decision.

On the opposite, user reviews data is similarly important for businesses to predict and monitor trends, study consumer attitudes as well as create directed action plans.

With more than 10 years of experience in gathering product reviews across an extensive range of e-commerce websites & product categories, X-Byte Enterprise Crawling is the go-to solution for all the data acquisition requirements.

Motivation

Due to their larger inventory, better customer service, and user experience, Amazon is always the leading retailer in a lot of countries. The majority of their products are reviewed well, with several in tens of thousands.

Using Machine Learning as well as Language Processing methods, we’ve studied lots of Amazon reviews. Our objective is to know consumer sentiments and assist in better decision-making.

Data Collection

data collection

For use cases, we have collected different reviews of exterior monitors from three top brands — Dell, HP as well as LG — in Amazon’s UK and US web stores.

Our dataset has incorporated 12 products from those three brands as well as the given data fields:

Product Details

  • ASIN
  • Brand
  • Price
  • Product’s Name
  • Seller

Reviews Details

  • Review Title
  • Complete Review Text
  • Helpful
  • Ratings
  • Review Date
  • Reviewer’s Name
  • Verified Purchase

After some rounds of QA (deduplication, removing null records, etc.), our last dataset included 10,786 records from which the first review was displayed in December 2014.

review over time

Top Brand Sharing with Similar Insights

top brand sharing with similar insights

Dell, HP as well as LG are the three most reliable brands in computers as well as the electronic accessory industry. As anticipated with the leading makes, the highest ratings of 5.0 had maximum reviews with 7,168.

By breaking down the given chart, we can observe that these brands have received nearly similar reviews in numbers overall as well as for every rating.

review by brands
total review by rating

These three brands had parallel average ratings also for monitoring on Amazon.

average rating by brands

Depending purely on numbers, one could gather that all the three brands have roughly similar market share in Amazon as well as is also as trendy as the other two. Future buyers would need more drills on review content to make the last decision.

Analysis of Top Reviewed Products

analysis of top reviewed products

HP 23.8-Inch VH240a Full HD Monitor was the maximum-reviewed product in the dataset having 3,559 reviews. So, we’ve chosen that to do individual analysis. This monitor also had average ratings of 4.27.

top review by product

100% Authenticated Purchases

For this selected monitor, nearly all the collected reviews were from authenticated purchases, imposing the quality of the dataset for more analysis.

verify purchase ratio

Average Quarterly Ratings over Time

The initial review for the HP VH240a monitor was given way back in July 2017. This achieved the maximum quarterly average ratings of 4.40 in the Q4 of 2019.

avg ratio over time

The average ratings have been dropping since the all-time high that might be down to some factors like:

Better options either within the similar brand or the competitor

New technologies in the display panel hardware as well as power management, picture quality, etc.

Happy Customers’ Ratio of 87%

With nearly two-thirds of reviewers (2,352) providing 5 stars, it’s quite fair to indicate that overall, the customers were pleased with the purchase. Opposed to it, merely 12% of the clients rated a monitor unwell with ratings either 1-2 stars.

review by ratings
helpful review per

Depending on the review percentage, which was supposed to be “helpful”, as given in the above chart, it is quite clear that with any higher percentage of users available, the negative reviews prove to be the most useful during the buying procedure. The percentage got calculated as:

Total reviews, which were marked useful * 100 / Total number of reviews for the rating

Although, while thinking about the real number of “likes”, total reviews of two excesses (ratings of 5.0 as well as 1.0) look to be most useful.

helpful review by ratings

Review Lengths

Depending on the analysis, we have found that the users have provided reviews of comparable lengths for every rating — about 350 to 400 words on an average. It suggests that, despite the observed product sentiment, the customers look to be equally descriptive about the experiences on reviews. It might be due to a fact that these monitors are a “well-versed buying product”, for which detailed data could be shared on the review.

Having seen that, there is a small increase in average lengths of reviews with every rating.

review lenght rating

Speakers, as well as audio quality, govern the customer opinions

common mention

Depending on a word cloud produced for good reviews, the customers are usually “happy” with the “quality” of the purchase, having “speakers”, “screen”, “sound”, as well as “picture” regularly mentioned. Customers also look to like monitors for “office”, “home”, and “gaming” objectives, as well as are ready to “suggest” that to the contacts.

bad review

Instead, dissatisfied customers look to be getting “problems” with “power”, “sound”, “cable”, as well as might “return” that.

As “speakers” featuring conspicuously in both the review groups, we could conclude that it is the feature, which is separating customers.

It was only a hint into the type of actionable analysis, which is possible with quality data, which X-Byte Enterprise Crawling offers. Analysts can do more thorough sentimental analysis for scraping extra insights from in-house or competitive review data that can be essential in driving well-versed product development decisions.

About X-Byte Enterprise Crawling

X-Byte Enterprise Crawling is the Web Data Scraping platform specialized in the withdrawal of web data from nearly all industry divisions. Contact us with all your requirements, and we provide a solution just for you.