Note that the most helpful review is someone who did not buy one. Same with the second. And the third.
> Sample provided for review.
> Got this at no cost for unbiased review.
> *I received a sample product from Ivation.
Reviews are much more scattered if you filter by "Verified purchase only". It's interesting that this drops the number of reviews from 165 to 104 for a product which is designed for and presumably sold only on Amazon.
I'm not necessarily suggesting these reviews are shady, but given the other aspects of Amazon that this company tracks, it would seem reasonable to suggest they've figured out a way to track influential reviewers likely to give them a positive review.
From what I can remember, Amazon has always been full of mostly useless reviews. Disclosing that products were provided for free does little to change that. About the only thing that does is let the parties involved "white wash" their hands of ethical issues.
Occasionally, there will be a shining star. Recently I ordered an AIO printer and a reviewer had a fairly comprehensive reviews on multiple models, almost at a short article length.
Music seems to be one category where I sometimes find value in the reviews. It seems like people are more forthcoming about their music tastes, even though subjective, than what they think about utilitarian products.
Does it bother anyone else that the three customer reviews on the product page all give five stars, and all three reviewers received the product for free in exchange for an 'unbiased' review? (And that fact is only visible if you click each 'see more' link at the end of the truncated review). I just have a hard time believing this practice doesn't create bias, even if unintentional.
I'm finding I have to work harder to parse meaning out of Amazon reviews than I did a few years ago. I don't see how this benefits Amazon, but I also don't see them doing much to fix it.
Some companies will provide the product "for free" but then -- and here's the kicker -- require the product to be returned or they'll charge you for it. I can realistically see an unbiased review coming from that process, but not one where the reviewer gets to keep the product for free.
It is typical in the "Private Label Business" to generate the first reviews using this technique. And Amazon only allows this in their TOS if the product was provided for free.
I would be interested to build a tool which tries to analyze the reviews e.g. using Natural Language Processing (NLP) or Machine Learning techniques to get the "real" picture for a given product. NLP could help to identify the most criticized aspects for a given product so a potential buyer does not have to go through any single review.
I am not an expert in NLP/ML though. Any ideas how this could be done?
tedmiston|10 years ago
> Sample provided for review.
> Got this at no cost for unbiased review.
> *I received a sample product from Ivation.
Reviews are much more scattered if you filter by "Verified purchase only". It's interesting that this drops the number of reviews from 165 to 104 for a product which is designed for and presumably sold only on Amazon.
I'm not necessarily suggesting these reviews are shady, but given the other aspects of Amazon that this company tracks, it would seem reasonable to suggest they've figured out a way to track influential reviewers likely to give them a positive review.
themartorana|10 years ago
ashmud|10 years ago
Occasionally, there will be a shining star. Recently I ordered an AIO printer and a reviewer had a fairly comprehensive reviews on multiple models, almost at a short article length.
Music seems to be one category where I sometimes find value in the reviews. It seems like people are more forthcoming about their music tastes, even though subjective, than what they think about utilitarian products.
guelo|10 years ago
EDIT: downvoters, care to explain what's wrong with this comment?
tomsaffell|10 years ago
I'm finding I have to work harder to parse meaning out of Amazon reviews than I did a few years ago. I don't see how this benefits Amazon, but I also don't see them doing much to fix it.
adventured|10 years ago
http://money.cnn.com/2015/10/18/technology/amazon-lawsuit-fa...
Wingman4l7|10 years ago
kiliancgn|10 years ago
I would be interested to build a tool which tries to analyze the reviews e.g. using Natural Language Processing (NLP) or Machine Learning techniques to get the "real" picture for a given product. NLP could help to identify the most criticized aspects for a given product so a potential buyer does not have to go through any single review.
I am not an expert in NLP/ML though. Any ideas how this could be done?
ck2|10 years ago
They are mostly bogus these days.
Unless you can find 3 star reviews, I usually read those.
muddi900|10 years ago