Most review sites unbox a product, wear it for a week and give it four stars. That tells you about one unit, in one pair of ears, on one week. It does not tell you whether the company will honour a refund, whether the FDA has ever heard of the product, or whether the 4,000 five-star reviews were written by the same three people.

Sound Judgment is run by a data scientist and professional evaluator. The method is evaluation first: pull every record that exists about a product and its seller, run it through the same checks every time, and reduce it to a short answer. Hands-on testing is reserved for products that pass the screen and matter enough to deserve it. Every article says which level of evidence it reached.

Level 1: The records

Before anything else, the paperwork. This alone disqualifies most of the market.

• Regulatory status. Is the device listed with the FDA as a hearing aid, or is it a personal sound amplifier being sold as one? Since the 2022 over-the-counter hearing aid rule this is where most of the deception lives. We check the listing, not the marketing.
• Specifications versus claims. The manual, the packaging, the FCC filing and the advertisement rarely agree. We note where they diverge.
• Return, refund and warranty terms. The real window, who pays return shipping, restocking fees, and whether the company behind the warranty has a verifiable address and phone number.
• Price history. Whether the “regular price” has ever been charged, and whether the countdown timer resets when you reload the page.
• Review pattern analysis. Rating distribution, review velocity, duplicated text, verified versus unverified purchases, and reviews that appear the same day a product launches. Fake review campaigns leave fingerprints, and this is where a data background pays off.
• Complaint records. Better Business Bureau files, FTC actions, recall databases, and the pattern of reader reports we receive.

Level 2: Cross-source synthesis

For products that survive Level 1, we pull together everything independent that exists: audiology research on the device class, published lab measurements where they exist, professional fitting guidance, and reports from readers with different types of hearing loss. The goal is to answer the question a single tester cannot: who does this actually work for, and who is it wrong for?

Level 3: Hands-on

Some products deserve to be worn. When we test hands-on, we buy at retail with our own money where possible, use the product for at least two weeks in ordinary life, and say plainly what type and degree of hearing loss the tester has, because a device that suits mild high-frequency loss can be useless for someone else. If a manufacturer supplied the unit, the article says so in the first paragraph, with no conditions accepted on what we publish.

What every article tells you

• The evidence level it reached: Records, Cross-source, or Hands-on.
• The date it was last checked. Products, firmware, prices and companies change.
• How the product was obtained, if it was obtained at all.
• The short answer first, then the evidence for readers who want it.

What puts a product on the Watchlist

A product or seller goes on the Watchlist when we can document at least one of these: a hearing aid claim without an FDA listing, a refund policy that is not honoured, fake pricing or fake scarcity, fabricated reviews or credentials, or a verified pattern of reader complaints. Each Case File shows the exhibits and the date they were checked, and the company is invited to respond in full.

What we will not do

We do not accept payment for evaluations, we do not let an affiliate relationship touch a verdict, and we do not pretend a records review is a wear test. If you know something has changed, tell us.