UX Breakdown of Customer Reviews
This is the age of ecommerce where we can buy anything from a hair pin to a house online. Online shopping has many advantages as compared to offline shopping, one of them being able to see what other buyers of the product are saying i.e. customer reviews and ratings. You can read about some interesting insights from customer ratings on Flipkart, Amazon and Snapdeal in my previous article.
While ratings are a short indication of how good or bad a product is, reviews are more detailed feedback of the product. They should ideally help you take informed decisions whether to buy a product or not; but are they?
Lets find out.
Let us analyze customer reviews for Moto E on Flipkart. Moto E has an overall rating of 4 out of 5 which gives you a feeling that its a good smartphone… but… for whom?
Here are some negative reviews about Moto E.
As we can see, most complaints are about low battery life, long time to charge and hanging issues. Imagine you are a person who travels a lot and are looking for a simple budget smartphone. Low battery life or taking long time to charge are deal breakers for you. This phone just doesn’t work for you if these are the problems people are facing. However, you may not be aware of this unless you read many reviews. If you just look at the specifications and overall rating, this looks like a great phone and you may go ahead and purchase it, which in your case, will be a wrong decision.
Here are some Positive reviews about Moto E on Flipkart.
As you can see, all these positive reviews talk about the budget. Yes, this phone has great specs at this price range. If you are looking for a cheap smartphone and don’t mind it being a little slow, this is a great choice.
Some other interesting insights from these reviews:
- Some people who have given 4 and 5 star ratings also wrote about cons, which means there are some bad things about the phone which they are ok with, as pros outweigh cons for them.
- Few people even suggested other phones which are comparable to this phone.
Essentially, the whole point is, I should be able to know what is good and what is bad about a product before making a purchase decision; because what is bad for others may not necessarily be bad for me; what is good for others may not be a major concern for me. Each individual is different and has different needs. We are shelling out our hard earned money to buy these products. We have a right to take the right decision!
How do we solve this problem?
Tripadvisor asks people to rate hotels on different parameters as shown in this screenshot. So, if I am a traveler who is most bothered about location, I may go ahead with this hotel even if it scores a little less in other areas. This can be replicated for ecommerce sites where people can rate a product on certain pre-defined parameters for that product category. For instance, a mobile phone can be rated on Performance, Battery, Features, Camera, Storage and Value for Money. However, this may not be a scalable solution because of the wide range of products and product categories.
Imagine if we can perform Natural Language Processing (NLP) and pluck out positive stuff from happy customers and negative stuff from unhappy customers and show them on the ratings, it will help us decide what features of the product made people happy and what made them sad.
It can also be used to suggest other products and do better cross selling.
One can even explore a different navigation pattern for showing a list of products based on customer preferences. This is particularly useful for people who don’t have a specific product in mind but just have a rough idea of what they want in the product.
It can also be used to create buying guides for cameras, speakers, ACs etc. Obviously, there is so much more that we can do with NLP on customer reviews.
To extend this further, lets say you selected a phone but you want to know more about battery life and heating issues. So, you may want to read reviews of people on these particular attributes of the product to take a call.
I don’t represent Flipkart / Amazon / Snapdeal in any manner. I also hardly have any knowledge on NLP. I am writing this only because nothing really is being done to address this problem in the ecommerce sites that I shop at regularly. Hoping for a better online shopping experience!
I thank Hemant Sagar for giving me some gyan on NLP & Sentiment Analysis.
Food for thought:
One of the most influential factors in our buying decisions is the opinions of our friends and relatives. Likewise, a…www.smashingmagazine.com
I don't know how Amazon's algorithm works. One simple scheme you could use to pick out the snippets in the example you…www.quora.com
Take free online classes from 80+ top universities and organizations. Coursera is a social entrepreneurship company…class.coursera.org
This is a summary of research paper as part of Social Media Analysis 10-802, Fall 2012. M. Hu and B. Liu. Mining…curtis.ml.cmu.edu