/ Amazon

How Amazon Can Accept More Honest Book Reviews Without Inviting Spam

Jeff Bezos and the Amazon team pride themselves on customer obsession: the customer always comes first. For the past two decades, they have largely lived up that promise. Amazon has the lowest prices, the most painless and speedy delivery, and a catalog literally from A to Z.

That said, Amazon can greatly improve how they deal with book reviews while staying true to their pro-customer philosophy. In this post, we propose an idea which they can implement to increase the number of honest book reviews without inviting spam. Our idea rests on the assumption that more honest book reviews is always better for the customer than less honest book reviews, all else being equal.

At the moment, two Amazon policies curtail the number of honest reviews a book can receive:

To post a review, customers must spend at least $50.00 using a valid credit or debit card. Prime subscriptions and promotional discounts don’t qualify towards the $50.00 minimum. Customers in the same household cannot submit a review for the same product.

This will reduce the number of honest reviews as many honest readers don't meet this requirement.

Customers can now only submit a limited number of non-Amazon Verified Purchase reviews a week. The limit is five and the count is calculated from Sunday at 12:00am UTC through Saturday 11:59 pm UTC.

This, too, will also reduce the number of honest reviews because many readers get free copies of books from non-Amazon sources (ex. ARCs from mailing lists). Some of these readers easily read more than 5 books a week, especially for shorter books.

We suspect the driving force behind both these policies is to combat review spam. The first one deters spammers since most spammers won't meet that requirement. The second one does damage control in case a spammer passes the first requirement. Neither policy gets at the heart of the problem, which is to validate that the review is honest. There are three pre-requisites for a review to be honest:

  1. The reviewer has the book.
  2. The reviewer read the book.
  3. The reviewer was not influenced.

For sake of argument, lets assume these books are ebooks. The first criteria can be met with the use of watermarks. The idea is not novel. Lets say an author wants to distribute 100 copies of her ebook for free. She could go to Amazon, generate a 100 watermarked copies of her ebook, each with a unique code. Then she could distribute them such that each reader gets their own uniquely coded copy. The reader can upload this watermarked copy on Amazon upon leaving the review. Amazon can cross validate the code with their internal records and, if there is a match, they know that the review is coming from someone who has the book (pre-requisite #1).

Amazon can institute safegaurds in place to combat spam with this approach: It can ensure that each code is used at most once to leave a review. So, 100 copies can generate max 100 reviews. Amazon could also invalidate the first review if it sees the same code twice from a different customer. This reduces the risk of fraudulent reviews from pirated copies and deters the author from spraying the same copy to many readers, praying for a single review. Furthermore, Amazon can impose a limit on how many watermarked copies an author can request; this limit could be a function of the author's historical reputation with Amazon and popularity.

As software engineers ourselves, we can confidently claim that the operational expenses for such a system will be a small dent in Amazon's bottom line. The implementation itself is not architecturally complex, especially for a 600,000 person company that is the brainchild of Amazon Web Services.

If Amazon builds this though, ebook distribution services like BookFunnel, CreateSpace, WattPad, etc... will have to update their systems to integrate seamlessly with these Amazon codes. Ideally, Amazon would expose an API to make this integration painless. It shoud not be difficult and it is worth every minute of an engineer's time, in our opinion; a significant number of honest reviews are at stake and, more importantly, Amazon buyer happiness is at stake.

With this system in place, the aforementioned policies can be amended to allow for more honest reviews without inviting the same level of spam; even if this brings in two spammy reviews for every ten honest reviews, we consider it a win:

Reviews for ebooks and audiobooks which pass the watermark test are not subject to the $50 or the 5 unverified reviews / week rule.

This kind of system will be especially useful for authors' ARC campaigns. ARCs already get special friendly treatment from Amazon, so they realize the importance of ARCs to the publishing industry. The above idea is congruent with that belief and we hope it is in cards.