How We Test and Verify

We do two things on this site.

We test gum — buy it, chew it, time it, score it.

And we verify who's selling it — because a significant share of the remineralizing and functional gum sold online is being sold by operators using another company's brand name, photographs, and product claims.

This page explains both, and sets out exactly what evidence we require before we publish a

finding about a seller.

1. How we get the products

Most products we review are bought at retail with our own money, from the same places you'd buy them: the brand's own store, Amazon, Etsy, TikTok Shop, or a shelf.

Every article states at the top how we obtained the product. There are four possibilities and we always name which one applies:

We bought it. Purchased at retail on a stated date from a stated seller. We keep the receipt.
It was sent to us free. We say which brand sent it. A free sample buys a disclosure, not a score.
The brand paid for queue priority. A brand can pay to be reviewed sooner. That buys scheduling only. We publish the result whatever it is, and the payment is disclosed at the top.
We haven't tested it. Some articles are research summaries — public reviews, ingredient records, seller documentation — where we have not chewed the product. These say so plainly and never carry a Tested Rating.

We do not accept payment in exchange for a rating, a ranking, or the removal of a finding. See How We Make Money for everything we do accept payment for.

2. Two kinds of rating, never blended

Tested Rating — only appears on products we physically tested. Scored 1–5, weighted:

Community Rating — a weighted average of public customer reviews across retail platforms: each platform's average multiplied by its review count, divided by total reviews across all platforms. We always state which platforms we included and the date we pulled them.

These measure different things and we never merge them into a single number. A product can carry a Community Rating and no Tested Rating, and the article will say so.

What we measure when we test

Flavor duration. We chew one piece and start a stopwatch, recording peak flavor (how long it holds full strength) and total flavor (when we can no longer detect it). Repeated across [3] pieces, reported as the median.

Texture and chew. Scored 1–10 on initial hardness, time to soften, cohesion versus crumbling, and whether it sticks to teeth or dental work.

Cost per piece. Price actually paid, including unavoidable shipping, divided by piece count. Reported in cents.

Ingredients. We photograph the actual panel on the pack we received and list every ingredient. We flag the gum base type and whether "plastic-free" claims match the label; every sweetener, and xylitol's position in the ingredient order where a xylitol claim is made; any functional ingredient and its stated dose; and any discrepancy between the marketing and the panel.

Panel scores. At least [3] testers. We publish individual scores, not just the average, so you can see where we disagreed.

3. How we verify a seller

A large portion of the remineralizing gum market is sold by operators using another company's brand identity. This is the most common problem we find, and it is also the most documentable — impersonation leaves evidence.

Our verification work is about establishing facts, not forming impressions. Every finding we publish is tied to a specific piece of evidence a reader can check for themselves.

3.1 What we document

For every seller we investigate we attempt to establish each of the following, and we publish which ones we completed and which we couldn't.

Image provenance. We reverse-image search the listing's product photography. Where an image originates with another company, we establish which company published it first and when, using the Internet Archive to timestamp the original. A listing running another brand's photographs is the clearest single indicator of impersonation.

Brand identity use. Whether the listing uses another company's name, logo, packaging design, product name, or marketing copy. We quote the copy side by side and link both sources.

Trademark records. Whether the brand being imitated holds a registered mark, with registration number and date. We link the public record.

Authorization status. Whether the seller is an authorized reseller of the brand whose identity appears on the listing. Where a trademark holder states on the record that a seller is unauthorized, we publish that statement and attribute it. Where we have no such statement, we say so — the absence of confirmation is reported, not glossed over.

Business entity. Whether a registered entity exists behind the listing, in what jurisdiction, and since when. We link the registry record.

Domain history. Registration date, and first appearance in the Internet Archive and in third-party traffic data.

Seller identity. Whether the marketplace seller name matches the brand on the packaging, and whether one seller account is running multiple brand identities.

Fulfilment reality. Where the product actually ships from, versus where the listing says.

Product-to-listing match. We order it. We photograph what arrives next to the listing, and compare the ingredient panel to the advertised formula.

Checkout conduct. Whether the checkout enrolls the buyer in a subscription without clear consent.

Responsiveness. Whether the seller answers a direct customer-service enquiry, and how.

3.2 The proof we require before we publish a claim

We hold each type of claim to a stated evidence standard, and we don't publish the claim without the proof.

We archive everything. Every screenshot carries a timestamp, and we submit the source page to the Internet Archive so the evidence survives if the seller edits their listing. We keep order confirmations, shipping labels, and correspondence.

3.3 Our verdict tiers

We don't use the word "scam" as a shortcut. Each verdict names what we actually established.

The line between "Multiple risk indicators" and "Documented brand impersonation" is the one we take most seriously. The first says a listing looks wrong. The second says we can prove whose identity is being used and when it was taken.

A small seller with a stock photo and slow shipping is not the same thing as an operator running another company's product photography under that company's name. We will not let the first be mistaken for the second, and we would rather leave a listing at "Unverified" than assign an impersonation finding we can't fully evidence.

3.4 Right of reply

Before we publish any verdict below "Established," we contact the seller at every address they publish and allow [7] days.

We state in the article that we contacted them, on what date, and whether they replied. If they reply, we publish their response. If they later provide evidence that changes the picture — proof of authorization, a corrected listing, a registry record we missed — we update the article and date the update at the top.

3.5 What a verification finding is not

Not a legal determination of fraud, counterfeiting, or trademark infringement. Those are for courts and regulators. We report documented facts about brand identity, business records, and what arrived in the box.
Not a judgment about the product's safety. We don't lab-test. A listing can be impersonating a brand and still contain gum.
Not permanent. Sellers change, listings change, and authorization status changes. Every finding carries a date and the evidence it rests on.

4. What our ratings are not

Not medical advice. We report what's on the label and what published research says about those ingredients. We're not your dentist.
Not a safety certification. We don't test for contaminants, heavy metals, or purity. When we say a panel lists something, we mean the panel lists it.
Not permanent. Formulations, prices and sellers change. Every review carries a "last tested" date. If it's old, treat it as old.
Not a consensus. They're our panel's scores, published individually so you can see the spread.

5. When we get it wrong

We will get things wrong. When we do:

We correct the article and add a dated correction note at the top. We do not silently edit.
If the error was material — a wrong verdict, a wrong ingredient, a wrong price — we say so plainly in the note.
If a company believes we've published something inaccurate, contact us with the evidence. We read every one, and we've updated articles on that basis before.
If we assign a verdict we can no longer support, we remove it and say why. A retraction we publish ourselves is worth more than one we're pushed into.