Yosef Bernath, 34, ran a company called Premium Home Service out of West Ridge, Chicago. He built more than 15,000 fake business listings, each with a local phone number, a nearby address, and a five-star rating. The consumer who checked the reviews found exactly what Bernath wanted him to find.

Since at least 2018, those listings showed up on search engines as small local shops: plumbers, electricians, HVAC techs. Each one looked like a real business down the road. None of them employed a single service technician. The consumer who typed "plumber near me" and picked the top-rated result got routed to a call center. The work, when it showed up at all, was done by unlicensed workers who often left the home in worse shape than they found it.

The Department of Justice and the FTC filed a complaint against Bernath and Premium Home Service in the spring of 2026. Employees and family members posted five-star reviews on command. Those reviews diluted the real one-star complaints, pushing the overall rating back up each time a genuine complaint appeared.

The system was built to catch one specific type of person: the one who follows the standard advice. Check the reviews. Pick the highest-rated local provider. Hire with confidence.

That advice sounds like due diligence. It is not. It is reading a vendor's marketing material and calling it research.

There is a name for this in the literature. Researchers at the University of South Florida published a study in Information Systems Research in 2025. Dezhi Yin, along with Samuel Bond at Georgia Tech and Han Zhang, ran five experiments over five years. They gave participants a set of reviews and asked them to sort each one as real or fake. The finding was consistent across all five rounds: consumers default to believing a review is genuine unless hit with strong contrary proof. The researchers call it truth bias. Your brain treats the five-star review as true until something forces it to reconsider.

That is not a trust problem. It is a wiring problem.

The exploit does not need a careless consumer. It needs a normal one. A person who reads a few reviews, sees the stars line up, and moves forward. That is how most people hire. The USF team found that asking users to report suspicious content is largely useless. Most users never flag what they cannot spot. The system relies on a detection skill that almost no one has.

FTC Commissioner Mark Meador said it plainly in his statement on the Bernath case. The consumers who tried hardest to find a strong local provider were the ones most reliably sent to Bernath's operation. Not the lazy searchers. Not the ones who picked the first name on the list. The diligent ones. The more careful the search, the more likely the consumer landed on a profile built to catch exactly that kind of search.

The platitude does not protect the careful buyer. It paints a target on him.

Here is the one thing every number above rests on: no review platform can verify whether the person posting is a real customer, an employee, or a relative posting on command. That is the gap. Everything else flows from it.

Google removed over 292 million policy-violating reviews in 2025 alone. That number is not a fluke. It is the size of the hole between what the platform shows you and what it can actually confirm.

The Transparency Company puts the total cost of review fraud at roughly $300 billion a year across home services, legal, and medical sectors. The average American household loses $2,385 a year by being steered toward vendors whose ratings were built, not earned.

Sit with that number. $2,385 per household, per year. Not from impulse buys gone wrong. From hiring the wrong plumber, the wrong lawyer, the wrong doctor, because the stars said they were safe.

The better principle: audit the reviewer, not the rating. Once that is clear, three moves follow from it.

Skip the fives. Go straight to the ones and twos. Look for patterns, not lone complaints. One angry review means nothing. Three reviews naming the same failure, late arrivals, hidden fees, unlicensed workers, that is a signal. The five-star reviews are where the marketing lives. The low-star reviews are where the operation shows its real face.

This takes more time than scanning a star count. That is the cost of a real check.

Click the name. On most platforms, you can see the reviewer's other posts. A profile with one review, five stars, and no photo is not proof of fraud by itself. But a cluster of single-review profiles all praising the same company in the same week is a pattern worth seeing. Real customers leave trails across months and categories. Manufactured profiles show up once and vanish.

Look up the company's state license number. Call the listed number and ask for the name of the technician they plan to send. Ask for proof of insurance. A real operator answers these questions in under a minute. A routing operation like Bernath's cannot. The license check takes less time than reading ten reviews. And it confirms the one thing stars never will: whether the business is real.

Running these three checks on your next few hires does something the star rating never did:

You see which vendors have real complaint patterns buried under managed ratings. You see which reviewer profiles were made for a single purpose. You see which companies answer a basic license question without a pause, and which ones stall, deflect, or go quiet.

After your next few hires, ask three things.

→ Which check gave you the most useful signal before the work started?

→ Which vendor looked strong on the rating page but fell apart under a simple question?

→ Which red flag showed up more than once across different searches?

That is the gap between advice that sounds right and a process that proves itself.

The consumer who searched "plumber near me" in Chicago found one of Bernath's 15,000 phantom listings. The five-star rating told him exactly what the vendor wanted him to hear. That is what star ratings are built to do.