Acting on product feedback is essential to optimizing your fit into the market. Your Amazon reviews provide you with surface-level information, but with the right analysis tools, you can cut through to truly meaningful feedback.
- Feedback overload
- A case study in Amazon reviews.
- What the breakdown of your Amazon reviews hides.
- Text analysis is the key to understanding.
- How you can extract top Amazon reviews.
Complaining about having too much data has become quite taboo in the current day and age, yet it is often seen as a good problem to have. However, it is a pain point for many marketplace businesses, who have to contend with a deluge of feedback, constantly arriving through reviews, customer service interactions and in comments sections. Processing this manually is incredibly technical, time-consuming, and prone to bias. This feedback in and of itself doesn’t say too much, but look deeper and you’ll uncover all sorts of insights.
What businesses need is a quick, accurate and consistent method for trawling and categorising this data, without having to splash the cash on technical workers like data scientists or engineers. Enter VAYU, an end-to-end data discovery tool that enables anyone to apply one-click artificial intelligence to a data set, saving time, resources and money.
A case study in Amazon reviews
Smartwatches form a segment of the wearable tech industry that is seeing a surge in popularity. The biggest players in the market are posting year-on-year growth in sales of over 100% as Fitbit struggle to keep up.
A crucial problem for Fitbit is that cheap alternatives are penetrating the market with incredulously low-price points and respectable functionality. Where Apple markets its premium Apple Watch Series 5 at over £400, consumers can pick up a plethora of smartwatches vastly homogenous to £150 Fitbits each for less than £50.
However, these review statistics hide what is really happening under the surface.
Studying 3 smart watches at 3 distinct price points (Apple Watch Series 5 – £429, Fitbit Versa 2 – £199, Willful Smart Watch – £43.99) can help tear down the mask worn by the review breakdown.
What the breakdown of your Amazon reviews hides
On each Amazon product, you are presented with a frequency distribution of 1 to 5-star reviews and an average score. Whilst giving the impression of a comparable statistic, this data is woefully inadequate.
More, it’s woefully inaccurate.
Comparing the complete set of reviews with the Top 50 most helpful reviews, there is a significant discrepancy in the distribution of ratings. Looking at just the Top 50 reviews, the percentage of 5-star reviews on the Apple Watch Series 5 and the Fitbit Versa had were 19% and 44% lower respectively.
The Willful smartwatch tells an entirely reversed story. The low price point watch instead has 5-star reviews overrepresented in the top 50 helpful reviews.
Text analysis is the key to understanding
Without some analysis of the review text, the reviews are worthless. This is where self-serve machine learning analysis can save you hours in text mining.
Combining sentiment analysis with topic modelling allows you to not only categorise the reviews into the most frequently discussed topics, but also to then rank these on how positive or negative these topics are spoken of.
Both the Apple watch and the Willful watch reviews are overarchingly positive. Apple customers were impressed with the battery life and its ability to monitor health, whilst Willful customers were impressed with the looks and how good it really was, most likely in reference to the low price.
Customers were far less impressed with the Fitbit Versa 2. The main complaints were with the new Alexa function, the new contactless payment function and the time it takes to set up the watch for use.
Without access to analyzing these most helpful reviews, both customers and Fitbit themselves may be disillusioned with the 4.4-star average and assume the product is well-received. It is important to dig deeper and extract more than just surface-level data.
How you can extract top Amazon reviews
In order to extract the most helpful reviews for each product, I suggest using the Amazon Reviews Exporter Chrome extension. The free version of this tool allows extraction of the top 50 most helpful reviews, and exports these into a CSV that requires little cleaning up before importing into VAYU.
VAYU empowers you to perform real data science in moments, and with text analysis, you can extract meaning in minutes.
We are told today that data is everything, yet we then become distracted by the data that we can see. Behind all data that is visible is a wealth of information that needs extracting, and now there is a tool that can be everyone’s guiding light with Amazon reviews.