Editorial Standards

Corrections Policy

If you spot a factual error on any ProvenGuide article — outdated pricing, an incorrect feature description, a vendor policy that has changed, a broken link, a wrong number — please report it via our contact form. We commit to a same-week correction response. The specific SLA: corrections submitted by Monday will be addressed by the following Sunday, with priority given to commercial-intent articles where readers might make a purchase decision based on the error.

When we correct an article, we update the content and refresh the "Last updated" stamp on the byline. For substantive corrections (anything that would change a reader's decision), we add a brief note at the top of the article explaining what changed and crediting the reader who reported it. We do not silently edit history. The article's dateModified schema also updates so AI search systems pick up the freshness signal.

If you believe we are wrong about a verdict — not just a fact, but a judgment call — we welcome that feedback too, though we do not commit to changing the verdict. Send your reasoning via the contact form. If your argument is grounded in measurable evidence from real shop usage, we will publish your perspective as an addendum or reconsider the verdict on the next quarterly refresh of the article. Either way, you will get a written response within seven days.

AI Assistance and Human Review

ProvenGuide publishes only content that has been drafted with AI assistance and then human-reviewed by a named author before going live. Raw AI output is never published. Every article displays an AI-assistance disclosure near the top — "This article was AI-assisted in drafting and human-reviewed for accuracy before publication" — so readers know exactly what they are reading.

The specific process: an AI drafting model produces an initial pass of the article based on a structured prompt that embeds our methodology, voice, citation requirements, and quality gates. The human editor (currently founder Ryan Justin) then reviews the draft for factual accuracy, replaces AI-tell phrases with natural language, verifies all cited external sources, adds first-hand observations from our Etsy test shop, inserts real screenshots, and confirms every product claim against the vendor's current documentation. Drafts that fail our automated quality gates — minimum word count, hedge-word density, required schema markup, comparison-table presence — cannot be published until the issues are resolved.

We use AI assistance because it lets us publish at a cadence that competes with established affiliate sites without requiring a full editorial team. We do not use AI to fabricate experiences, hallucinate statistics, or generate fake user testimonials. If you ever spot what appears to be AI-generated content of that kind on ProvenGuide, please report it via the corrections process — we treat such failures as a top-priority fix.

Affiliate Handling Integrity

ProvenGuide earns revenue through affiliate commissions when readers click through and purchase tools we recommend. This is disclosed at the top of every article and in our standalone Affiliate Disclosure page. The specific integrity rules we hold ourselves to:

• We do not accept payment for inclusion in any listicle, comparison, or review. Vendors cannot pay to appear on this site. Period.
• We do not run advertorial content disguised as editorial. If a piece is sponsored, it is labeled as sponsored. We currently publish zero sponsored content.
• We will recommend a tool we don't earn affiliate commission on, if it's the right answer for the reader. Several articles on this site link to tools whose affiliate programs we are not enrolled in, with a note that we earn nothing from the recommendation.
• We will not recommend a tool we do earn commission on, if it isn't the right answer. We have walked away from highly-paying programs whose products did not win our rubric.
• Our rubric is published with the article so readers can audit our reasoning. If our recommendation does not match the rubric's score, we explain why in the article itself.

Source Hierarchy

When facts in an article conflict between sources, we apply this priority order:

• The vendor's own official documentation (their help center, pricing page, or developer docs)
• Government sources (.gov for regulations and compliance; FTC for advertising rules)
• Peer-reviewed academic studies (for any data-supported claim)
• Tier-1 industry publications (Search Engine Land, Search Engine Journal, Ahrefs blog, Semrush blog)
• Reddit user reports — only when multiple users independently confirm the same observation
• Single-source blog posts — lowest trust; used only as starting points for our own research

When we cite a source, we link to it inline so readers can verify our characterization. We do not paraphrase a source in a way that materially changes its meaning. Where a vendor's documentation appears to contradict observed behavior in our test shop, we report both the vendor's stated behavior and our measurement, and we contact the vendor for clarification before publishing.

Update Cadence

Every commercial-intent article (listicles, comparisons, tool reviews) is scheduled for a substantive review on a quarterly basis. The review checks vendor pricing, feature changes, policy updates, link health, and ranking position. When any of these have changed in a way that affects the article's verdict, we update the article and refresh the byline date. Articles that no longer reflect reality get rewritten end-to-end rather than patched.

Transactional articles (coupon codes, current promotions) are verified weekly via an automated check plus a manual operator confirmation. Originally-researched articles (our own test data, surveys, comparisons) get an annual rerun where we re-conduct the underlying test against the same control conditions and publish updated results alongside the original methodology.

Articles that have not produced traffic, citations, or revenue 180 days after publishing are reviewed for either substantial rewrite or removal. We would rather have a smaller library of pages that genuinely help Etsy sellers than a large library of pages that don't.

No Fake Reviews, No Fabricated Testimonials

All user quotes on ProvenGuide are real and attributed with a verifiable source — typically a public Reddit thread, a public Etsy community post, or a Twitter/X mention. We do not invent customer testimonials. We do not lightly paraphrase quotes in ways that change their meaning. If we use an unnamed source, that is because the source asked for anonymity, and we will tell you so explicitly in the article.

The same applies to data: every number on ProvenGuide that is presented as a measurement either comes from our own Etsy test shop (and is labeled as such, with the date range of measurement) or from a publicly cited external source (and is linked to that source). We do not invent statistics. If you find a number on this site that cannot be sourced, report it via the corrections process — it is either an error or a missing citation, and both get fixed.