This page explains who creates the content on Good Health Reviews, the rules every page follows, and exactly how we use AI.
Who’s behind the content
Good Health Reviews is run by Mauro Costa, a real person who tracks the health launches people are talking about. Mauro Costa is not a doctor, nutritionist, or other health professional, and our pages don’t claim otherwise.
We don’t currently have a medical reviewer. If that changes, the reviewer’s name, credentials, and review date will appear on every page they check.
Our editorial rules
Every page on the site follows these rules:
- No made-up experience. We never claim to have taken or used a product unless we really did. If we did, the page says so and shows our own photos. We never invent results, timelines, or personal stories.
- No social proof we can’t back up. No testimonials, quotes, customer counts, or star ratings that we can’t verify.
- No disease claims. We don’t say a supplement treats, cures, or prevents any disease.
- No false badges or promises. We never call a supplement “FDA approved” (the FDA doesn’t approve dietary supplements before they’re sold), we never promise results, and we never call a product risk-free.
- Claims are attributed. When we mention what a product is supposed to do, we make it clear that’s the seller’s claim (“the sales page says…”) and then look at what the research shows.
- Research is cited. Scientific statements link to the study on PubMed, and we describe what the study actually found, in whom, and at what dose.
- Facts come from the source. Prices, packages, and guarantee terms come from the official sales or order page, with the date we checked. If we can’t verify something, we say it’s not disclosed.
- Real pros and cons. Every review includes genuine downsides and a safety section.
- No pressure tactics. We don’t copy countdown timers, “only 3 bottles left” warnings, or similar urgency from sales pages.
Our tone is friendly and curious. We’re not here to trash products, but we won’t leave out something a buyer should know.
How we use AI
We use AI tools, currently Anthropic’s Claude, to research and write our reviews. Here’s how that works, step by step.
What AI helps with
- Reading the sales page and pulling out the claims, ingredients, doses, prices, and guarantee terms.
- Searching PubMed for studies on each ingredient and summarizing them.
- Searching for public complaints and customer feedback.
- Writing the review, following the rules above.
What our software checks automatically
- That every study cited really exists on PubMed.
- That the draft doesn’t contain phrases we never use, such as common disease-claim wording (“cures,” “treats,” “prevents,” or “reverses” a disease), “FDA approved,” or invented personal experiences. A pattern check can miss unusual wording, so if you spot something off, please tell us.
- That the basics are there, like real downsides and enough cited studies.
- That prices, cost per day and the money-back guarantee add up and match everywhere on the page.
- That the seller’s affiliate rules are followed (for example, no bonuses or coupons of our own) and that every paid link is labeled.
Software can confirm that a study exists. It can’t fully confirm that the study supports the sentence it’s cited in. To lower that risk, the abstract of each study is read during the research step before it’s cited, and the page says who was studied, at what dose and for how long, so you can check it yourself through the PubMed link.
How a page goes live
Reviews are published automatically. A new page goes live only if it passes every automated check above; if any check fails, it stays unpublished as a draft. No one reads each page before it goes live.
Mauro Costa sets the rules those checks enforce and corrects errors when they’re found, whether by readers or later by us. Every correction is noted on the page and on our Corrections page.
What AI never does
- AI is never listed as the author. Which product we cover comes from a fixed ranking (how much attention an offer gets on ClickBank and what it pays affiliates), and whether a page goes live is decided by fixed rule checks, not by the AI that wrote it.
- AI doesn’t invent experiences, testimonials, or numbers.
- AI doesn’t create product photos or “results” images. The header images on our reviews are AI-generated illustrations of ingredients, like herbs, roots or fruit. They’re decorative and never show the actual product, a person, or a before-and-after.
AI can make mistakes, and so can people. That’s why we link our sources and keep a Corrections page.
How often we publish
We publish up to two reviews a day. Each one goes through the same research steps and the same automated checks, and a page that fails them is not published.
Comparisons, ingredient guides and our data page
Besides reviews, we publish three other kinds of pages, all built from the data in reviews we’ve already published and from PubMed abstracts:
- Comparisons (“X vs Y”) of two products we reviewed in the same topic, only when people actually search for that comparison. The side-by-side table comes straight from both reviews. We never name a winner: we show where each one differs and who each may suit.
- Ingredient guides for ingredients that appear in several products we reviewed: what human studies found, at what doses, and how the label amounts compare. Ingredient guides have no paid links.
- Supplement Label Check, a data page that tallies, across every product we reviewed, how label doses compare with the doses used in studies, plus cost per day and refund terms. It’s generated automatically from our reviews and has no paid links.
These pages are written with AI assistance and published automatically when they pass the same automated checks as our reviews, plus extra ones: prices must match our reviews, and dose numbers are checked against our reviews and the cited studies. When a source review changes or comes down, the pages built on it are updated or taken down too.
Sellers and money
We earn commissions through affiliate links, mostly through ClickBank. Our Affiliate Disclosure explains how.
Sellers don’t pay us to cover their products, and they don’t see or approve pages before they go live. If a seller sends us an update, we check it against the source. We update facts that have changed, but we don’t remove accurate, sourced information on request.
Keeping pages up to date
Prices, formulas, and guarantees change. We re-check our most-read reviews about every three months, and we update any page when we learn something changed. We change the “last updated” date only for meaningful changes, like new prices, a changed formula, or a correction. Typo and broken-link fixes don’t change it, and we never change a date just to make a page look fresh.
Questions about this policy
Email contato@webflag.com.br. If you think a page breaks one of these rules, please tell us. We’d rather hear it from you.
Last updated: November 2026.