Editorial standards
These standards guide how Top10AIs prepares rankings, identifies commercial relationships, and handles changes. They apply to new work and substantive revisions.
Evidence before claims
Factual claims should be supported by sources that readers can inspect. Official documentation supports statements about a provider’s offering; public reviews describe the experiences of their authors. Neither should be presented as an independent Top10AIs performance test.
Unsupported performance promises, fabricated quotations, invented review consensus, and invented scores do not belong in a ranking. Limitations and contradictory evidence should be explained when they could affect a reader’s choice.
Editorial independence
A tool’s relevance, documented capabilities, practical limits, pricing, and available evidence guide its place in a comparison. Affiliate availability is not a ranking criterion.
Our policy does not allow payment, free access, or another commercial relationship to buy inclusion or a higher position in an editorial ranking. Any sponsored content must be clearly labeled and kept distinct from ranked recommendations. A material commercial relationship connected to an original test must be disclosed with that test.
Vendors may submit factual corrections through the same corrections process available to readers. A requested change needs supporting evidence; a commercial relationship is not a reason to remove a supported criticism.
Affiliate links
Top10AIs may earn a commission when a reader purchases through an affiliate link. A disclosure appears on ranking pages. The presence of a link is not a guarantee of quality, availability, or a particular outcome.
Use of AI
Top10AIs uses AI to assist with research, drafting, organization, and formatting. Sources are checked against the claims they support during preparation, but errors can still occur. AI assistance is not a substitute for supporting evidence, and generated text must not invent experience, quotations, or test results.
Featured illustrations may be AI-generated. They are decorative article artwork, not screenshots or evidence of a product’s output. Product logos identify the tools being discussed and do not imply endorsement by their owners.
Original examples and hands-on work
Original examples must distinguish observed output from explanation. A test note should identify the tool and plan when known, the date, relevant settings, the input, what happened, and any limits on the conclusion. Screenshots or examples supplied by a vendor must be labeled as such.
A small example can illustrate a workflow. It should not be turned into a broad claim about accuracy, conversion rates, reliability, or typical results without evidence that supports that claim.
Corrections
Readers and vendors can flag a possible error through the site’s contact form. Include the article URL, the disputed statement, the proposed correction, and a supporting source if available.
Corrections are assessed against the available evidence. When a confirmed error materially affects a recommendation or a reader’s understanding, the article should be corrected and the change noted. A submitted report is not a guarantee that a requested change will be made.
Dates and meaningful updates
A research date identifies when the stated evidence was checked. A publication date identifies when the article first appeared. A change to the site design or a routine copy edit should not be represented as a new research review.
Substantive update notes should describe the actual change—for example, revised pricing, a changed plan requirement, a discontinued product, or a replacement in the list. Dates must not be refreshed simply to make content appear newer.