Editorial Standards

Research standards

Every review starts by identifying the product’s claimed ingredients and doses from the current sales page and, where available, a purchased or photographed Supplement Facts panel, not from marketing summaries alone.

• Databases searched: PubMed, PubMed Central (PMC), and the NIH Office of Dietary Supplements fact sheets are our primary sources for human evidence. We also check ClinicalTrials.gov for unpublished, ongoing, or completed trials of the specific finished product, and journal publisher sites directly (e.g. ScienceDirect, MDPI) when a citation needs verifying against its original text, not just an abstract.
• Evidence hierarchy: we grade evidence on an A–U scale, A (the finished formula itself was studied) down through B, C, D (mechanistic or animal-only) to U (unverifiable), explained in full on How We Grade Supplement Evidence. A single-ingredient human trial is never treated as equivalent to a trial of the actual product being reviewed.
• Study designs: randomized controlled trials carry the most weight; we still report observational studies, cohort data, and animal or in-vitro research when it’s the best evidence available, but we say plainly which kind it is, rather than letting a citation imply stronger proof than it is.
• Animal and laboratory research: reported as mechanistic support, never presented as evidence of an effect in people.
• Manufacturer-funded studies: disclosed as such where we can identify the funding source, and weighed with the same skepticism any peer-reviewed process would apply to a funder with a direct financial interest in the result.
• Dose and formulation matching: the core of our methodology. We compare the dose used in the cited research against the dose the product actually discloses (or, where a proprietary blend hides individual amounts, we say so explicitly). A real citation for an ingredient is not treated as support for a product that delivers a fraction of the studied amount.
• Date cutoff and update schedule: we don’t run reviews on a fixed monthly/quarterly cycle. Each review states the date its pricing, sales-page claims, and citations were last checked (see each post’s Update Log). We re-check a review when we’re re-notified of it, via reader correction, a rescan, or a scheduled content pass, not on a guaranteed calendar interval.

Corrections standards

Anyone can report a factual error, a broken link, or an outdated claim to smileysocial@digiprospark.com.

• Minor corrections (a typo, a dead link, a stale price) are fixed directly, without a separate public note.
• Material corrections (a factual claim that turns out to be wrong, a citation that was misattributed, a mislabeled ingredient dose) are fixed in the article text AND logged in that post’s Update Log with the date and a plain description of what changed. We do this even when the correction is unflattering to the original version, which is the actual point of keeping a log.
• Score changes: if a correction changes any of the five Trust Score pillars, the recalculated score is shown along with the reason in the Update Log, never silently adjusted.
• Formula changes: supplement formulas do change over time. When we learn a product’s formula has changed, we re-verify the current Supplement Facts before updating the review, and note the change and the date we caught it.
• Retractions: if a review turns out to rest on a source that was itself retracted or fundamentally wrong, we retract the affected claim and note it. We do not quietly delete the error and republish as if it never happened.

Product selection

• How products are chosen: primarily from established affiliate networks (ClickBank, Digistore24, and comparable programs) covering categories we already research: weight management, blood sugar, brain & focus, oral & dental health, energy, and immune support. A product being available through an affiliate program is what makes it eligible for review, not evidence that it will score well.
• Affiliate availability and selection: yes, affiliate availability affects which products we’re able to review at all. We can’t review or link to a product with no working affiliate or purchase pathway. It does not affect the score a reviewed product receives; see our Affiliate Disclosure for the full separation between compensation and verdict, and the FTC’s own endorsement guidance for the disclosure standard we follow.
• Can brands request a review? A brand or its representative may suggest a product for review via Contact. A request does not guarantee coverage, a timeline, or any particular outcome.
• Free products: we do not solicit free products in exchange for a review, and receiving a free product from a brand would be disclosed on that review if it ever happened. As of this page’s last update, every reviewed product was purchased or evaluated from its publicly available sales page and Supplement Facts listing, not a complimentary sample.
• Can brands preview a review before publication? No. We do not share draft scores, verdicts, or article text with a brand before publishing, and a brand cannot pay to change a score, add a claim, or suppress a Conditional or Not Recommended verdict.

Artificial intelligence policy

We use AI tools as part of how this site is researched and built, and we’d rather say so plainly than have a reader assume otherwise from generic disclaimer language. Specifically:

• Research discovery: AI tools are used to search and cross-reference published literature (PubMed, PMC, journal publishers) alongside our own manual searches, not as a replacement for reading the actual source.
• Drafting: AI assists in drafting review text, comparison tables, and FAQ answers from the ingredient evidence, dosing data, and citations gathered during research.
• Summarization: AI is used to summarize longer source material (a full clinical trial, a systematic review) into the shorter form used on a review page, with the summary checked against the source before publication.
• Image generation: we do not use AI-generated imagery to depict a specific product, its packaging, or its Supplement Facts panel. Product and label photography is either the vendor’s own official material or, where noted, not available. Generic illustrative graphics (icons, diagrams) may be AI-assisted.
• Editing: AI is used in the editing pass: checking internal consistency (e.g. that a stated dose matches the ingredient table elsewhere on the same page), tightening prose, and catching the kind of drift that happens across a large, growing site.
• Schema: the structured data (JSON-LD) markup on this site is generated programmatically by the theme’s own code from the same fields a human editor fills in, not separately AI-authored, and never asserting a rating or fact that isn’t also visible on the page itself.
• Translation: this site does not currently publish in any language other than English; no translation workflow, AI or otherwise, is in use.

Human checks before publication: every citation is checked to confirm it links to a real, findable source (not a fabricated or misattributed one); every stated dose is checked against the product’s own disclosed label or third-party retail listing, with the source flagged when we couldn’t independently verify it; every Trust Score is recomputed from its five stated sub-scores to confirm the arithmetic matches what’s displayed; and every page is read in full by DigiproSpark’s founder, Satwasheel Gharde, before it goes live. AI drafts; a human is responsible for what publishes.

Medical review policy

• Which pages would receive priority for professional review, if engaged: blood-sugar and diabetes-medication interaction content, autoimmune and immunosuppressant warnings, tinnitus and hearing-related claims, cognitive/nootropic ingredients such as Huperzine A, nail-fungus and differential-diagnosis content, and pregnancy/nursing cautions: the pages where an inaccurate safety claim carries the most real-world risk.
• Scope, should a reviewer be engaged: a medical reviewer’s role would be to check safety-adjacent claims (interactions, contraindications, dosing cautions) for clinical accuracy, not to re-score or endorse a product.
• Difference between editorial and medical review: every published page already goes through editorial review: citation-checking, dose-verification, arithmetic-checking, and a full human read, described above. Medical review is a distinct, additional clinical-accuracy check that has not yet been added to that process.
• How you’d know if a page ever gets one: a genuinely medically-reviewed page would carry a visible “Safety reviewed by” credit with the reviewer’s real name and the date reviewed, built into the review page itself and ready to switch on the moment a real reviewer relationship exists. It stays off until then. We will not fabricate a reviewer credit to make a page look more credentialed than it is.

Security & responsible disclosure

This site doesn’t collect payment details, passwords, or sensitive personal data directly: checkout happens on the vendor’s own site, and the newsletter/contact forms only ever ask for an email address and, optionally, a message. If you find a genuine security issue anyway (a vulnerability in the site itself, not a complaint about a third-party vendor’s checkout page), we want to know about it before anyone else does.

• Security contact: smileysocial@digiprospark.com, the same address used for corrections and general contact. A machine-readable version of this contact is also published at /.well-known/security.txt per RFC 9116.
• Responsible disclosure: please report a genuine vulnerability privately, by email, before disclosing it publicly, and give us a reasonable window to fix it. We won’t pursue legal action against a good-faith report made this way.
• What we don’t need reported: the affiliate-cloaking redirect links behave as designed (short URL → vendor’s real checkout); that’s not a vulnerability. Third-party vendor checkout issues should go to the vendor, not to us.
• Transport security: HTTPS is enforced sitewide with HSTS (including subdomains, preload-eligible), and standard security headers (Content-Security-Policy, X-Content-Type-Options, X-Frame-Options, Referrer-Policy, Permissions-Policy) are set on every response. This section was last reviewed against the live site’s actual response headers on 2 August 2026.