The best reviews come from bribes: if 40% of online reviews are fake, aren’t we just paying for fraud?
At the end of last year, a PC magazine published data showing that 39% of online reviews are fake. Other studies have shown that 61% of Amazon’s 3C product reviews are fake, 63% of beauty product reviews are fake, and 59% of sports shoe reviews are fake… Out of all these fakes, only one thing is true: online reviews are completely unreliable. But most people don’t know this, or even if they do, they still can’t avoid being deceived.
We’ve all had this experience: seeing a product we like in a physical store, picking it up and looking at it from all angles, but still hesitating to buy it… How we wish there were reviews on the shelf – like Amazon’s rich and convenient reviews – to help us make a decision. We’re no longer used to buying products without reviews.
Maybe you should think about whether having reviews is really better for you?
Trust is the beginning of fraud
Reviews are the most important mechanism for online shopping. Without reviews, there is no trust, and without trust, online shopping cannot exist. Why should we pay a stranger for a product we can’t see or touch? The buying and selling mechanism that humans have built for thousands of years relies not on trust, but entirely on seeing is believing.
But in the internet age, a different mechanism must be established.
The earliest online reviews can be traced back to the founding of eBay in 1995. They believed that the foundation of the company’s survival was to establish mutual trust between people, so they had to have a publicly available evaluation system. eBay’s early motto was “People Are All Good”. This slogan was meant to make people trust the mechanism.
The only thing that made me dare to click the mouse to place an order when I first shopped on eBay was the seller’s thousands of 100% positive reviews. At that moment, the seller’s score was more important than the price. If thousands of people had already bought from him without a negative review, then I should be able to place an order with confidence. Like most people, my online shopping began with trust in the review system. Today, eBay has been around for 25 years, and the whole world is still using this mechanism.
But today, the trust behind it has completely deteriorated.
When Ratings Become Assets
One of the key reasons for Amazon’s success today is that they have turned ratings into rich product information. With tens of thousands of ratings for many of their products, Amazon is the only place in the world where people can get a deep understanding of a product. These ratings are Amazon’s assets and their cash cow, so they do everything they can to encourage buyers to leave ratings in various ways.
Initially, they only allowed people who had actually purchased the product on Amazon to leave a rating. Then they opened it up to people who had not purchased the product. The idea behind this was that consumers who had purchased the same product elsewhere might still be willing to share their experience on Amazon.
This free asset is extremely valuable to them, but it has also opened up a path for fake ratings.
The World of Fake Ratings
When the pandemic hit and the entire United States was in lockdown, online businesses were thriving. This sudden surge in demand also stimulated the growth of fake ratings. The CEO of Fakespot, in an interview with CNBC, stated that Amazon’s fake rating rate has always been around 30%, but has increased to between 35% and 40% since the outbreak. So nearly 40% of Amazon’s ratings are fake, which is the same as the average for e-commerce sites reported by this PC magazine.
This is Amazon, a super enterprise with sufficient assets, capabilities, and expertise to catch fake ratings, and yet they still have this problem. The prevalence of fake ratings on other websites can only be imagined.
The severity of this problem is that 95% of consumers refer to online ratings before clicking the mouse, and 85% of consumers believe in online ratings. However, 40% of what consumers believe is fake. Research by e-commerce consultant John LeBaron shows that for every additional star in a product’s rating, sales increase by 26%. With such a high return, businesses naturally do everything they can to improve their ratings.
Deception with High Returns
Merchants can work hard to improve quality, please picky customers, and fight an honest battle… but they can also avoid the trouble because the best ratings come from “bribery”.
Buying fake ratings with money is not illegal, but the effect is immediate.
Most online shoppers only look at two numbers: the star rating (quality) and the total number of ratings (quantity). On Amazon, if the rating is good, the number of ratings is high, and it is marked with the “Amazon’s Choice” label, orders for that product will flood in. Trust in online shopping is more important than price, especially for high-priced products. A survey by Northwestern University found that sales of high-priced products with these three combinations could increase by 380%. The algorithm for Amazon’s Choice is not clear to the public, but the most important indicator is the rating. If the product has a high rating, the probability of being upgraded to Amazon’s Choice is greatly increased. Once endorsed by Amazon, consumers will no longer be wary.
However, consumers do not know that even selected products are still full of fake ratings.
Although Amazon’s Choice has very strict standards, the higher the standard, the higher the risk. Amazon’s artificial intelligence also has “fake rating factories”. In addition, due to the large scale of counterfeiting, it is difficult for Amazon to completely eliminate these obstacles when facing 600 million products and 5 million merchants.
The Underground Economy of Fake Ratings
To make fake ratings good, you need to know the techniques. There is a systematic and progressive model behind it. There are even advertising companies on the Internet that design models for merchants and outsource execution to professional personnel. Behind this is a complete product chain that produces fake ratings day and night. Anything that is counterfeited in the world may have legal risks, including making fake news, except for fake ratings. The only cost of accidentally making a fortune is being delisted by Amazon and never doing business again. Under such “completely unequal rewards and punishments”, it is no wonder that so many people are willing to take this risk.
There are many types of fake ratings, from machine “canned ratings” to “sweatshop click factories”, to the highest-end “professional fake rating hands who actually buy, actually unbox, actually take photos, and write a real unboxing article with emotional content”. The cost varies depending on the weight of the influence.
Making fake ratings requires patience and skill, especially before the product is just launched and the reputation has not yet been formed. It is necessary to cooperate with the actual number of buyers to make fake ratings, otherwise it is easy to be caught by Amazon’s AI. Usually, on average, only one out of every thousand consumers is willing to leave a rating, so merchants must follow the principle of proportion and cooperate with the rhythm. Secondly, the ratings that consumers trust the most are between 4 and 4.7. Perfect 5-star ratings often make people suspicious, so balance must be maintained in the score, and occasionally some neutral ratings should be made.
The Double-Edged Sword of AI
While machine-generated comments can easily be caught and removed by Amazon’s AI, the advancement of AI technology is a double-edged sword. As Amazon’s AI becomes more accurate, fake reviews also become more convincing. In the past, machines could only write canned phrases that were obviously fake, such as “A++”. Now, with the help of publicly available AI technology, machines can leave comments that resemble those written by real people, allowing them to evade Amazon’s AI police.
There are even free “fake review generators” available online, so you don’t even need to spend money to commit fraud. Simply enter the product category, and the generator will immediately produce AI-generated reviews. If anyone is still using canned reviews from ten years ago, they are not keeping up with the times. However, fake review generators can only be used for small-scale deception. To be effective, you need a large number of fake accounts, which is where fake review factories come in.
The Sweatshops of Fake Reviews
These “click factories” are mostly located in India, Bangladesh, or Vietnam. Contractors use local underground organizations to produce fake reviews day and night using a large number of fake accounts. If you are willing to spend money, you can find cheap labor overseas through intermediaries to engage in fake review production… They earn only 25 cents in US dollars for each review they produce. Because the reviews are customized by humans, it is difficult to determine whether they are real or fake unless there are fixed patterns of grammar and spelling errors.
If you need more convincing and sophisticated fake reviews, you can also use websites like fiverr.com, which connects you with anonymous freelancers for a fee of $5 to write customized high-end reviews using fake accounts. In addition, businesses can purchase fake reviews from specific Facebook groups for an average of $6 per review. There are more than 20 such groups on Facebook, with an average membership of 16,000 people. This is a unique labor market on Facebook alone.
The Most Honest Counterfeit Goods
However, so far, all of these fake reviews are not from actual buyers. They are far less significant than the reviews left by Amazon-verified buyers in terms of point calculation… If you think ahead, the merchants have also thought of it – that is, spend more money to find someone to actually buy, actually unbox, actually upload a bunch of photos, and really post a touching unboxing article. This is the most authentic and weighty fake review, and it is also impossible to be caught because “everything is real except for their inner thoughts.”
These people appear to be enthusiastic Amazon loyal customers who spend a lot of money on Amazon shopping and rely on writing the most authentic fake reviews to receive gifts or rebates from merchants. Making a set of services with Amazon “verified buyers” can earn an average of $25. If the price of the product is not high, after the matter is done, the merchant will refund the money through other online payment channels, treat the product as a gift, and add a small reward.
Professional Fake Reviewers
A technology media has tracked such a full-time fake reviewer. Miss Jessica has evaluated more than 700 products, and there are piles of Amazon packages waiting to be evaluated in every corner of her home. It is not so easy to engage in this industry because merchants will provide a series of instructions for buyers to follow. Amazon tracks all customer clicks, and Jessica cannot directly enter a specific product to place an order. She must first enter some keywords provided by the manufacturer to deceive and slowly browse… just like a real buyer. After receiving the product, she deliberately waits for five or six days before leaving a review, indicating that it has gone through a real trial period. These step-by-step details are to ensure the authenticity of the review, and this is just one example reported by the media. There are actually countless similar examples.
On average, Amazon adds 50 new products from China every second. It is difficult for new products without brands and reputation to sell on Amazon. Investing a little money to give away free products in exchange for fake reviews is the fastest way to establish a sales channel. Once the reviews become popular, it can stimulate sales. Once the sales increase, the product will also rise in the priority order of the search. Starting is always difficult, and the difficulty lies in establishing reviews. These Chinese merchants who participate in bribery will first use free small products to fish on public Facebook groups. Once they find specific targets, they will transfer the conversation to private groups and establish a long-term fraudulent relationship.
Spies never tire of deceiving each other
Sellers who boast about their products and have clear targets are easy to catch, but it’s hard to use AI to leave negative reviews for competitors. This is because leaving reviews only has one chance and the source is often unknown. This means that deception is spread out. If you see a bunch of one-star reviews with vague comments like “it broke after two uses,” it’s likely that these are negative reviews given by competitors. But this may only be the first step, as there are even more malicious tactics to come.
Usually, Amazon will put bad reviews at the end, as a single negative review is not enough to change the overall rating. Enemies will hire people or use machines to click “this review was helpful” to increase the visibility of the negative review and make it appear at the top of the ratings. These clicks are anonymous and do not leave comments, so as long as the IP source is different, Amazon has no reason to stop them. Therefore, leaving a negative review is just the seed, and the clicks that follow are the watering and growth.
Leaving negative reviews for enemies may sound a bit malicious. But even more malicious is leaving a large number of 5-star reviews for enemies.
As mentioned earlier, if the content of the positive reviews is empty and frequent, and the pace does not match the sales rhythm, it is easy for AI to catch them and lower their search priority, making it difficult for the world to find them, and even leading to removal. These 5-star gifts that fall from the sky all come from overseas underground factories, making it impossible for you to figure out who is so thoughtful. So if you are a seller and suddenly have a lot of nonsensical 5-star reviews, don’t be too happy too soon.
How can consumers protect themselves?
In the world of e-commerce, where there are many scams, there are also tools to catch them. The website I use most often is Fakespot.com. Just enter the product’s URL and the filtered truth will be revealed.
4.5 stars becomes 1.5 stars
Last year, I was interested in a folding bike on Amazon with a rating of 4.5 stars. After analyzing it with Fakespot, the true rating dropped to 1.5 stars, classified as a D grade. The analysis showed that 48% of the reviews were fake, and the effective rating was only 1.5 stars on average. This spring, during the height of the pandemic, I found a blood oxygen monitor with 7.6K reviews and a rating of 4.5 stars. However, after filtering it with Fakespot, it was downgraded to 3 stars and classified as a C grade because 26% of the reviews were fake.
4.5-star rating on Amazon
The same product with only 1.5 stars after being filtered by Fakespot
Fakespot’s secret to catching fake reviews is their business secret, but they have also publicly revealed some basic rules, such as unusual words from the same region, unusual sentence structures or errors, and unverified buyers who do not follow the proportion principle… etc. Their tool for catching fake reviews is nothing more than using AI and machine learning. Both of the products I mentioned above were listed as “featured” on Amazon.
Therefore, consumers are starting to question why Fakespot can catch fake reviews, but Amazon cannot. Is it because their technology is inferior? Is the scale of cheating too large? Or is catching fake reviews not a priority for their business interests?
Fake Reviews = False Advertising
If a physical store sells products that do not match the reality, regardless of where the products come from, they have legal responsibility. However, when e-commerce companies encounter legal issues, they lower themselves to “just providing platform services” and can only try to be fair and just, but cannot guarantee that there are no loopholes. Therefore, they should not be punished together.
Their analogy is that if someone publishes false advertising in a newspaper, the business that publishes the advertisement is committing a crime, not the newspaper. Of course, we all know that the biggest difference here is that the newspaper did not intervene in the transaction, but purely acted as a platform. We can also clearly see two problems: first, e-commerce intervenes in collecting money, processing orders, and shipping, but lowers itself to a platform, so it does not need to bear legal responsibility. Second, fake reviews are not equivalent to false advertising in the eyes of the law, so they have always been in a gray area, which is the main reason why they are so rampant.
At the end of last year, the US Federal Trade Commission finally took the first step and filed a “false advertising” lawsuit against a weight loss drug seller who did fake reviews on Amazon, which led to a consumer’s kidney failure. Although Amazon was spared, this also informed the world that fake reviews can be treated as false advertising, and future online fake reviews may also use this clause.
In addition, the California State Assembly will also propose a clause that requires e-commerce platforms to bear the same legal responsibility as physical stores. If this law is passed, Amazon’s ability to catch ghosts will surely be foolproof overnight. However, as consumers, we should also know that if we blindly trust reviews without doing our homework, we are essentially feeding fraudsters with our money.
Before clicking the mouse to place an order next time, please think carefully about how much of the review you rely on is true.