00:00:00
From counterfeit goods to fake N-95
masks, price gouging to disappearing
00:00:05
orders, shoppers on Amazon have a
growing need to proceed with caution
00:00:08
before clicking Buy Now.
00:00:10
Since Amazon's early days, reviews are the
one big metric customers rely on
00:00:14
to determine the quality and
authenticity of a product.
00:00:18
Turns out many of those
reviews can't be trusted.
00:00:20
The review system as
of today is broken.
00:00:24
Before the pandemic, the usual benchmark
around our average fake reviews
00:00:29
was 30%.
00:00:30
The norm has now become
close to 35%, 40%.
00:00:34
In recent years, thousands of fake
reviews have flooded Amazon and Walmart,
00:00:38
eBay and others, just as
sales numbers have skyrocketed.
00:00:42
And as shoppers stay home, online orders
are up 57% since the same time
00:00:46
last year and the number
of reviews is up 76%.
00:00:50
There's an element where you simply want
to trust those stars and you want
00:00:53
to trust the numbers, because if you
can't trust that, how do you know
00:00:57
what you're buying? From Facebook groups
where bad actors solicit paid
00:01:01
positive reviews to bots and click
farms that upvote negative reviews to
00:01:05
take out the competition, fake reviews
have boosted sales of unsafe
00:01:09
products, caused huge brands to sever
ties with Amazon and hurt business
00:01:13
for legitimate sellers.
00:01:15
We can't compete. We can't surface our
products that are new and innovative
00:01:19
and truly valuable to consumers because
other products that aren't so
00:01:23
great are playing this
game of review manipulation.
00:01:27
We decided to find out why
fake reviews have infiltrated Amazon, how
00:01:30
customers can spot an unreliable review,
and what the trillion dollar
00:01:34
company and others are
doing to stop them.
00:01:44
One big draw over competitors like
Walmart, Target and eBay is that
00:01:48
Amazon's listings often have hundreds or
even thousands of reviews instead
00:01:51
of just a handful. It's so easy, no
matter what site you're on, to simply
00:01:56
say the most reviews with the most
stars means the most level of
00:02:00
happiness. It's just simply
not the case.
00:02:02
If those Amazon customers aren't really
customers or if they're an
00:02:06
organization of paid individuals who just sit
there and go five star, five
00:02:11
star, five star, that doesn't really
tell me anything meaningful about the
00:02:15
product. Review software company Bazaarvoice
did a study of 10,000
00:02:18
consumers at the end of last year.
00:02:20
42% of consumers are saying that fake
reviews from the brand itself would
00:02:27
cause them to lose trust.
00:02:28
82% of those consumers are saying that
would cause them to never buy that
00:02:32
brand again. The problem is fake and
real reviews are getting harder to
00:02:36
tell apart. When you have no reason
to think it's a fake review, that's
00:02:39
when the consumer's in
the most danger.
00:02:41
And as shoppers increasingly turn online
for things they'd normally want to
00:02:45
shop for in person, like the nursing
bras made by Simple Wishes, there's a
00:02:48
higher chance of serious repercussions from
the purchase of a counterfeit
00:02:52
or low quality product.
00:02:53
And if the product's Amazon page
is filled with fake positive reviews,
00:02:57
shoppers won't know to steer clear.
00:02:59
We see reviews of people saying that
their breast tissue was torn and
00:03:03
irritated and bleeding because
of irritating seams.
00:03:07
And, you know, we see things like this
or like this product broke or it
00:03:11
tore after I wore it three times.
00:03:14
You see those real reviews surface and
then all of a sudden there'll just
00:03:17
be massive positive reviews.
00:03:19
A high rating can also trigger
the coveted Amazon's Choice badge, although
00:03:23
Amazon did say it will delete the
badge if a product isn't adhering to
00:03:26
policy. Amazon prohibits any attempt to
manipulate reviews and told CNBC
00:03:31
it will suspend, ban and take legal
action against those who violate these
00:03:35
policies. For any review, even the most
genuine, it always is worth asking
00:03:41
why is someone writing that review?
00:03:43
What is the incentive
to write that review?
00:03:45
Free products and payment
are increasingly common incentives.
00:03:48
Sellers solicit pay-for-play reviews through
popular Twitter accounts and
00:03:52
Facebook groups with
thousands of members.
00:03:54
So I joined some of these groups
just to kind of poke around.
00:03:57
And the first groups I joined, there
were five different postings from our
00:04:02
competitor asking for a review.
00:04:05
I felt like I just struck gold
finding my competitor there, reported it to
00:04:10
Amazon and nothing happened.
00:04:12
UCLA and USC released a study in July
that found more than 20 fake review
00:04:17
related Facebook groups with an
average of 16,000 members.
00:04:21
In more than 560 postings each day,
sellers offered a refund or payment
00:04:26
for a positive review,
usually around $6.
00:04:29
Amazon told CNBC it works with social
media sites to report bad actors who
00:04:33
are cultivating abusive reviews
outside our store.
00:04:36
And we've sued thousands of bad actors
for attempting to abuse our reviews
00:04:39
systems. The FTC requires reviewers to
disclose any payment or connection
00:04:44
to the product being reviewed.
00:04:45
On some sites like Fiverr and
Freelancer, users get around this by
00:04:48
advertising marketing services, a thinly
veiled reference to pay-for-play
00:04:52
reviews. There's also the more direct
approach where sellers include a
00:04:55
note inside a package asking for a
review in exchange for a discount or
00:04:59
other compensation. It's hard to keep on
top of five million sellers and
00:05:04
600 million products.
00:05:05
There's always a few bad seeds in the
mix, and it's the bad seeds that get
00:05:09
the attention. It's not that
Amazon's sitting back doing nothing.
00:05:12
It's that the scope of what
we're dealing with is so vast.
00:05:15
There are legitimate paid reviewer
programs like Amazon Vine, Early
00:05:19
Reviewer and Amazon Associates, which
require reviewers to disclose that
00:05:23
they've received a product for free in
exchange for what's supposed to be
00:05:25
an honest review. But Amazon has little
way to detect a compensated review
00:05:29
when deals are made
outside these programs.
00:05:32
There's a Velcro panel in the back so
you can constantly reset the size and
00:05:35
it's always the proper support.
00:05:37
Sisters Joy Kosak and Debra Abbaszadeh
designed a new type of hands-free
00:05:41
pumping bra and started selling it on
Amazon in 2009, where sales took off
00:05:45
quickly. But for the past three years,
sales have been flat, dropping off
00:05:49
after Amazon started to openly court
Chinese sellers to join its
00:05:52
marketplace. Cheaper bras with an
exceptionally similar design to theirs
00:05:56
started popping up, getting hundreds
of five star ratings seemingly
00:05:59
overnight. When that happened, we saw
a pretty immediate race to the
00:06:04
bottom in terms of pricing.
00:06:07
The sisters have been tracking review
activity on listings from competitors
00:06:10
like Momcozy and sharing
the data with Amazon.
00:06:12
Our best seller, where we used to be
number 25 in baby, we over the past
00:06:20
ten years of being on Amazon, we
have collected a little more than 10,000
00:06:24
reviews. It took them a couple of
months to to increase by 4,000.
00:06:32
Big brands like Nike and Birkenstock
have been so burned by competitors
00:06:35
selling knockoffs with thousands of five
star reviews that they stopped
00:06:38
selling on Amazon altogether.
00:06:40
Although Nike's landing page still appears
active on Amazon, the items
00:06:43
there are being sold
by third-party sellers.
00:06:46
They're fake, they're counterfeit.
00:06:47
They're either bought from Alibaba or
eBay and then they're resold on
00:06:51
Amazon. So a lot of these
sellers are actually ruining Nike's reputation
00:06:56
and they're putting in all the reviews
into the official listing for Nike.
00:07:01
At times, big brands themselves
are soliciting fake reviews.
00:07:04
Last year, for example, skincare brand
Sunday Riley settled with the FTC
00:07:08
after it was caught encouraging employees
to post fake reviews on
00:07:11
Sephora.com. On Amazon if you're not doing
some sort of, you know, tricky
00:07:15
technique, it's at least one hundred orders
for each review that you get.
00:07:20
Bernie Thompson sells about 120
consumer electronics products on Amazon
00:07:24
from his warehouse outside Seattle.
00:07:26
Competitors have tried to undermine his
sales with fake review tactics.
00:07:29
We've had people take our most
negative review, the one that's most
00:07:33
embarrassing, and we've had competitors
vote up those negative reviews.
00:07:36
Let's say your competitor has a one-star
review on the first page, you can
00:07:40
buy 100 helpful votes.
00:07:42
When they're considered most helpful, they show
up at the top of the
00:07:45
results. And so you can really
harm your competitors by doing that.
00:07:48
That helpful box can easily be clipped
by bots instead of humans or by
00:07:52
click farms overseas.
00:07:53
The ones that I've been contacted by
are all in Bangladesh, India, I think
00:07:56
one of them, Vietnam.
00:07:58
They have computers and they've got
fake accounts and they basically turned
00:08:02
in this whole system where they go in
and just click on "helpful" once and
00:08:06
then log into a different account and
then click on "helpful" again and so
00:08:09
on to where you can just pay
for basically taking down your competitors.
00:08:13
Bots are also getting better
at generating convincing written reviews.
00:08:17
We actually see a lot of these
fake review farms leveraging open source
00:08:20
projects from these behemoths, such
as Google, Open AI, multibillion
00:08:25
dollar research firms and leveraging
it to produce fraud.
00:08:29
And by this case, we're producing human
like text that looks like really
00:08:34
realistic. Amazon's own algorithms do
usually detect these patterns and
00:08:38
remove them within weeks.
00:08:39
Amazon says we're going to wait 30
days and if we detect that there's
00:08:44
enough fake reviews, we'll pull
back those fake reviews.
00:08:47
The problem is, during that 30
day policing period, the product can
00:08:51
generate a whole lot of sales
that it didn't otherwise deserve.
00:08:54
In 2019, Amazon changed its review system
so customers can leave a simple
00:08:58
star rating with one click instead
of a full written review.
00:09:01
This tool that Amazon put out there to
make it easier for consumers to give
00:09:05
real feedback has actually made it
easier for the scammers to elevate
00:09:10
their star rating, just the volume, because now
all they have to do is say
00:09:14
all you have to do is click a button.
00:09:16
No one can tell who left the rating.
00:09:18
You will not see those ratings as a
list of authors on the bottom of the
00:09:22
page. And we see products with thousands
of ratings that have no body,
00:09:26
text body attached to them.
00:09:28
While a rating can only be left
by someone who bought the product, Amazon
00:09:32
allows reviews from anyone even if
they haven't made a purchase.
00:09:35
We see certain categories have over 90%
of the reviews on the product are
00:09:40
unverified. And when you look at them, it
just looks like a flood of bot
00:09:44
reviews. What Amazon does is they
give different weights to different
00:09:47
kinds of reviews and so a verified
purchase review will have more of a
00:09:50
weight than someone
who wasn't verified.
00:09:52
But the intention is that you could have
bought it at Walmart and want to
00:09:55
review it. You could have bought it
somewhere else and want to review it.
00:09:59
And then there's a slew of new
tricks popping up from bogus seller accounts
00:10:02
to mysterious free Amazon packages
appearing on people's doorsteps.
00:10:06
In one tactic known as Review Highjacking,
a seller takes over a once
00:10:09
popular listing. So you'd have these
crazy situations where, you know, our
00:10:14
product was a USB hub but
we had to discontinue it.
00:10:18
And somebody's selling like women's eyelashes
would take over that product,
00:10:22
change the picture to women's
eyelashes, change all the text.
00:10:26
The reviews would show
these 2,000 positive reviews.
00:10:29
But if you'd read the
reviews, they're not about eyelashes.
00:10:32
They're about a USB hub.
00:10:34
Another recent tactic involves seed packets
from China showing up at
00:10:37
hundreds of people's houses who don't
know where they came from.
00:10:41
The Better Business Bureau warns that
the scam, often called Brushing,
00:10:44
means the seller is using the seeds
to generate fake Amazon orders tied to
00:10:47
U.S. addresses. Then they can write
fake verified reviews about themselves
00:10:51
falsely inflating their
seller rating.
00:10:53
Then there's sock puppet reviews, which
are bogus accounts created by a
00:10:57
seller to write positive reviews
on their own products.
00:10:59
Sellers can also hack into a customer's
Amazon account and post a positive
00:11:03
review from there without
the customer ever knowing.
00:11:06
And they're all new products that are
getting reviews at an amazing rate.
00:11:09
It's just not, it's not believable.
00:11:12
With so many ways to create
realistic fake reviews, some start-ups have
00:11:16
developed ways to detect them.
00:11:18
Fakespot is one of these.
00:11:19
Fakespot launched a new Chrome plugin in
May that has a quarter million
00:11:22
downloads so far. It analyzes the
credibility of a listing's reviews and
00:11:26
gives it a grade from A to F.
00:11:28
The Fakespot Guard will actually catch
these sellers dynamically as you're
00:11:31
browsing Amazon.
00:11:33
And we will offer you an alternative
seller that is authentic and genuine
00:11:38
that we've seen before that
has high customer satisfaction.
00:11:40
Other online tools that customers can
use to check the credibility of
00:11:44
Amazon reviews include ReconBob, ReviewMeta,
the Review Index and Review
00:11:48
Skeptic. Shoppers willing to spend time
to vet their purchases can
00:11:52
manually spot fake reviews, too.
00:11:54
The number one way consumers tell us
they identify a fake review are
00:11:57
multiple reviews with the
same language in them.
00:12:00
So they're basically looking for
patterns in the reviews.
00:12:03
The second most important way is reviews
that are not actually about the
00:12:07
product. The third is
poor grammar and misspellings.
00:12:12
And the fourth, and I actually think this
is one of the more important ones
00:12:15
is overwhelming number of
five-star positive reviews.
00:12:18
If a product only has two or three
reviews that it's gathered over a long
00:12:24
period of time and those two or
three reviews look pretty good, consumers
00:12:29
actually need to give kind of more
trust to a product like that.
00:12:32
Clearly, that brand and that
manufacturer, they're not gaming anything.
00:12:36
If you do spot a fake review,
Amazon encourages customers to use the report
00:12:40
button next to each review.
00:12:41
But whether Amazon will take any action
after fake reviews are reported is
00:12:45
a different question. We go down these rabbit
hole that take a lot of time
00:12:49
to look for this information and then
we share it with Amazon and nothing
00:12:52
happens. And it's just exhausting.
00:12:54
After CNBC brought Simple Wishes' complaints
to Amazon, months after it was
00:12:58
first informed of the illegitimate
reviews, Amazon said, "We've taken
00:13:01
appropriate action on these accounts." Amazon
told CNBC it uses powerful
00:13:06
machine learning tools and skilled
investigators to analyze over ten
00:13:09
million review submissions weekly, aiming
to stop abusive reviews before
00:13:12
they're ever published. Getting Amazon
to actually do investigations,
00:13:16
quite frankly, they don't have enough
investigators to do all the possible
00:13:19
investigations needed.
00:13:21
When I was at Amazon, there was
a time when Amazon had about 20
00:13:24
investigators for the
whole United States.
00:13:26
There was over a million sellers on Amazon
at the time and there were 20
00:13:29
investigators. In an unprecedented move,
Amazon hosted a virtual
00:13:33
conference earlier this month to give tips
and listen to concerns from its
00:13:36
third party sellers, who make up
58% of Amazon's e-commerce business.
00:13:41
When it comes to outside regulation,
fake reviews are prohibited by the
00:13:44
FTC, but it's a complex issue.
00:13:45
Where you can leave a review and
you receive some kind of compensation, you
00:13:50
need to put in a disclaimer.
00:13:51
And that's consumer law.
00:13:53
That's been around for a while.
00:13:56
But there are different ways that
this is now being gamed.
00:13:58
There is no law attached to ratings
where you can leave them without text.
00:14:02
Targets and Walmart, they are they
are held to a higher standard.
00:14:08
They have to vet products that they
put on their shelves or through their
00:14:12
e-commerce platform because
they are liable.
00:14:15
And that's the huge difference here.
00:14:17
Unless Amazon is purchasing the product
from the seller as a wholesale
00:14:22
purchaser and they are representing as
the seller, they have zero
00:14:25
liability. And that's frightening.
00:14:29
Last year for the first time, the
FTC prosecuted a company for fake reviews
00:14:32
on Amazon. The inflated reviews were
for a weight loss supplement that's
00:14:36
made with a plant that
can cause acute liver failure.
00:14:39
You can already see the FTC
becoming more interested in reviews.
00:14:43
They treat reviews as a form of
advertising because of the influence that
00:14:48
it has on us as shoppers.
00:14:50
Now, Amazon supports a California Assembly
bill that would subject online
00:14:54
marketplaces to the same
product liability requirements as
00:14:57
brick-and-mortar retailers, despite years
of Amazon successfully fighting
00:15:00
lawsuits against such rulings.
00:15:02
If passed, it could incentivize Amazon
and others to better police fake
00:15:06
reviews. Amazon owns the keys to that
data, and they they can do it.
00:15:10
I know they can. As Amazon continues
to help people stay safely at home,
00:15:14
the need for shoppers to trust the
reviews and order with confidence has
00:15:17
never been higher. It's really almost a
societal level issue of, you know,
00:15:22
can Amazon kind of keep control of
its systems and live up to the
00:15:27
dependency that we have on them?
00:15:30
And I think the you know, honestly,
they've been growing so fast that
00:15:34
they've been struggling with it.
00:15:35
On one hand, Amazon is
getting better about policing.
00:15:38
On the other hand, it's
a cat and mouse game.
00:15:40
You know, that probably
will never end.
