Why Online Shopping Is Broken — And Who Profits From It

The Product That Arrived Looked Nothing Like the Ad

You've seen it happen, maybe even lived it. You order something online — a gadget, a piece of furniture, a piece of clothing — and what arrives looks like a rough sketch of what you actually purchased. The photos were polished, the reviews were glowing, the price felt reasonable. And yet the product in your hands is a disappointment at best, a scam at worst.

This isn't a fringe problem. It's a structural one baked into how online shopping works — and it's getting worse as artificial intelligence makes deception cheaper, faster, and harder to detect. Understanding the mechanics of why online shopping is broken isn't just useful for avoiding bad purchases. It's essential for anyone making financial decisions in a world where the gap between what's advertised and what's real has never been wider.

---

Information Asymmetry: The Oldest Trick in Commerce

Economists have a name for the core problem: information asymmetry. It describes any transaction where one party knows significantly more about the product than the other. Sellers almost always hold that advantage.

This dynamic is centuries old. The Latin phrase caveat emptor — "let the buyer beware" — has been in use for at least 400 years as an acknowledgment that buyers must do their own due diligence because sellers will not do it for them. Horse traders of the 17th century were notorious for obscuring an animal's age or health. Savvy buyers learned to inspect the horse's teeth — a reliable age indicator — before committing to a deal. Hence the phrase "don't look a gift horse in the mouth."

Misleading advertising didn't start with the internet either. Campbell's Soup famously placed glass marbles at the bottom of bowls to push ingredients to the surface for photography. Cereal brands used glue instead of milk so flakes looked perfectly crisp on screen. These tactics were legal then under the doctrine of puffery — the idea that a degree of exaggeration in advertising is expected and therefore not technically deceptive.

Courts have upheld this standard repeatedly. When Wendy's and McDonald's faced lawsuits over burger photography that bore little resemblance to the actual product, judges dismissed the cases on the grounds that no reasonable consumer would take promotional images at face value.

The uncomfortable logic: because we expect to be misled, we can't claim to be deceived.

---

How AI Has Industrialised the Online Shopping Scam

If puffery was a grey area before, AI has turned it into a free-for-all. The cost of generating a hyper-realistic product image — showing a robot that moves fluidly, a lamp that glows perfectly, a jacket that fits like it was tailored — has dropped to near zero. What used to require a professional photographer, a stylist, and a physical sample can now be produced in seconds with a text prompt.

The scheme follows a predictable playbook:

• Step 1: Use AI to generate a compelling, photorealistic product listing that grabs attention and looks credible.
• Step 2: Source cheap, low-quality physical goods from manufacturers willing to ship something that only vaguely resembles the promoted item.
• Step 3: Buy a layer of five-star reviews from professional brokers to establish social proof before real customers can leave honest feedback.
• Step 4: Once negative reviews accumulate and the listing tanks in search rankings, shut down the shop and reopen under a new name.

With Amazon hosting an estimated 600 million product listings, identifying and shutting down fraudulent sellers before they turn a profit is an almost impossible task at scale. By the time enforcement catches up, the margin has already been made.

The legal environment makes this worse. Under Section 230 of the Communications Decency Act of 1996, online platforms cannot be held liable for content posted by third parties. That means Amazon, Google, and every major e-commerce platform is legally insulated from responsibility for fake listings and fabricated reviews hosted on their sites. Regulators like the FTC can pursue the sellers and review brokers — but never the platforms themselves.

---

The Fake Review Economy: $150 Billion in Influenced Spending

User reviews were supposed to be the equaliser. If sellers hold an information advantage, surely thousands of honest consumer opinions would level the playing field?

In theory, yes. In practice, the review system has been comprehensively gamed.

According to the World Economic Forum, approximately 4% of online reviews are likely fake — and that figure is probably conservative, since many go undetected. These fraudulent reviews are estimated to influence more than $150 billion in global spending annually.

The mechanisms range from crude to sophisticated:

• Incentivised reviews: Real customers are offered free merchandise, refunds, or discounts in exchange for changing a negative review to a positive one.
• Review brokers: Professional services — many operating overseas, beyond the reach of US regulators — flood new listings with five-star ratings for a fee. Finding these brokers takes minutes; public Facebook groups openly recruit reviewers and offer payment.
• AI-generated text: Fake reviews used to be detectable by awkward grammar, generic phrasing, or suspicious enthusiasm for mundane products. AI has solved that problem. Modern synthetic reviews read as naturally as genuine ones and can be produced at industrial volume.

Big platforms are deploying AI-powered detection systems in response. But the historical pattern is not encouraging: in the cat-and-mouse game between fraud and detection, the fraudsters tend to adapt faster than the enforcers.

---

Why Platforms Are Conflicted — and Unlikely to Fix This Alone

Here's where the incentive structure becomes revealing. Amazon and similar mega-retailers have two competing interests:

1. Reputation: A platform known for selling junk loses consumer trust over time.
2. Revenue: Every sale — legitimate or fraudulent — generates a commission.

The short-term financial logic favours tolerance of the problem. As long as enough consumers keep buying, keep not returning items, and keep trusting reviews despite knowing better, the economics of fake listings remain viable. Platforms benefit from the transaction volume even as they publicly condemn the behaviour.

This is not unique to retail. Social media platforms have long understood that engagement — even outrage, even misinformation — drives time-on-platform, which drives ad revenue. The underlying incentive structure is identical: the platform profits whether or not the content is real.

Legal liability would change this calculus immediately. If Amazon could be sued for hosting a listing it knew or should have known was fraudulent, the financial motivation to invest in robust enforcement would be obvious. Section 230 removes that pressure entirely.

---

The Social Proof Trap — Even for High-Stakes Decisions

Perhaps the most troubling dimension of this problem is how far it extends beyond consumer goods. Psychologists describe social proof as the cognitive tendency to validate our decisions by observing what others have already done. It's why queues outside restaurants signal quality, why bestseller lists influence book purchases, and why five-star ratings feel reassuring even when we know they can be bought.

For a $20 phone case, the stakes of misplaced social proof are low. But consider:

• 75% of patients consult online reviews as their first step in finding a new doctor, according to survey data.
• Multiple independent studies have found virtually zero correlation between a medical professional's online rating and the actual quality of care they provide.
• The same review infrastructure — gameable, manipulable, and largely unregulated — is used to select contractors, therapists, legal professionals, and financial advisers.

The information asymmetry problem, in other words, isn't just about whether your lamp looks like the picture. It's about whether the surgeon you selected based on a 4.9-star rating deserves that rating at all.

---

What Savvy Shoppers and Investors Can Actually Do

The system has structural flaws that individual behaviour won't fix. But there are practical steps that meaningfully reduce your exposure:

For product purchases:

• Prioritise independent review outlets — Wirecutter, Consumer Reports, and similar publications whose only asset is credibility, not commission revenue.
• Use tools like Fakespot or ReviewMeta to analyse the authenticity of review patterns before purchasing.
• Take advantage of return policies aggressively. Returns are costly for fraudulent sellers and create data trails that platforms use to identify bad actors.
• Search for the product name alongside "Reddit" or "forum" — community discussions are harder to manipulate at scale than listing reviews.

For high-stakes decisions:

• For professionals (doctors, lawyers, contractors), treat online ratings as a starting filter, not a final verdict. Verify credentials independently through professional licensing boards.
• Ask for referrals from people you know personally. Word-of-mouth from a trusted contact has a accountability structure that no review platform can replicate.
• When evaluating financial products or advisers online, cross-reference with regulated databases — FINRA's BrokerCheck in the US, for example, provides disciplinary history that no amount of five-star reviews can obscure.

The uncomfortable baseline: 90% of shoppers check reviews before purchasing, even as awareness of fake reviews grows. Knowing a system is flawed does not automatically change behaviour. Building deliberate habits around information sourcing — defaulting to independent verification rather than platform-curated ratings — is the practical answer until regulation catches up.

---

The Structural Fix That Isn't Coming Soon

The most direct lever available to policymakers is revisiting Section 230 liability protections for e-commerce platforms in the context of demonstrably fraudulent listings. If platforms faced meaningful legal exposure for hosting products and reviews they had reasonable cause to flag as fake, the incentive to invest in enforcement would follow immediately.

The political will for that change remains limited. Section 230 has become a politically charged topic across multiple fronts — free speech, content moderation, platform power — and any amendment faces significant opposition from tech industry lobbying. In the absence of structural change, the burden falls disproportionately on consumers.

In the meantime, the playbook for fake listings will keep evolving. AI-generated imagery will become more realistic. Synthetic reviews will become indistinguishable from genuine ones. The gap between what is shown and what is sold will widen.

The economic concept of caveat emptor was coined in a world where buyers could physically inspect what they were purchasing. Four centuries later, it's being applied to a marketplace where the product might not physically exist at the time of listing. That's not a grey area. That's a broken system — and knowing how it works is the first step to navigating it.

---

This article is for informational purposes only and does not constitute financial advice. Always consult a qualified financial professional before making investment decisions.

Frequently Asked Questions

How widespread is the fake review problem in online shopping?

The World Economic Forum estimates that roughly 4% of online reviews are likely fake — a figure most researchers consider conservative given detection limitations. These manipulated reviews are estimated to influence more than $150 billion in global consumer spending annually. The problem spans major platforms including Amazon, Google Maps, and app stores.

Is it legal for sellers to use AI-generated images in product listings?

Currently, yes — with caveats. AI-generated imagery falls under the same advertising standards as traditional photography, meaning a degree of idealisation is permitted under the legal doctrine of puffery. However, if the image materially misrepresents the product in a way that causes consumer harm, it can cross into false advertising territory. Enforcement remains inconsistent, particularly for sellers operating across international jurisdictions.

Why don't platforms like Amazon do more to stop fake listings?

The incentive structure is conflicted. Platforms earn a commission on every sale — legitimate or not — and bear no legal liability for third-party listings under Section 230 of the Communications Decency Act. While reputational damage is a long-term concern, the short-term revenue from fraudulent transactions creates limited urgency for aggressive enforcement. Meaningful change would likely require platforms to face direct legal liability for hosting demonstrably fraudulent content.

How can I tell if online reviews are fake before making a purchase?

Several signals and tools help. Look for review clustering — many reviews posted within a short window, or an unusual ratio of five-star to one-star ratings with nothing in between. Tools like Fakespot and ReviewMeta algorithmically analyse review patterns and flag suspicious activity. Searching a product name on Reddit or consumer forums often surfaces unfiltered feedback that is significantly harder to manipulate than platform reviews. For high-value purchases, independent review outlets like Wirecutter or Consumer Reports — which operate on credibility rather than commission — are more reliable sources.

Does the fake review problem apply to services, not just products?

Yes — and this is arguably the more serious dimension. Survey data shows that 75% of patients use online reviews as their first step in selecting a new doctor. Multiple studies have found no statistically meaningful correlation between a healthcare provider's online rating and the quality of care they deliver. The same manipulable review infrastructure that inflates ratings for phone accessories is used to build the online profiles of contractors, therapists, legal professionals, and financial advisers.