Show the evidence.
Keep the limits visible.
How we review sources, explain product differences, and maintain our articles. Policy updated October 21, 2026.
1. Label the evidence
Our guides link product claims to the relevant documentation. Readers can follow the sources to check scope and current details. Editorial recommendations connect those facts with the requirements of a specific creative task.
2. Prefer traceable, scoped sources
Use official product documentation, model references, pricing pages, and applicable provider terms for product-specific facts. Link the exact relevant page where possible. A checked date records when a source was examined, not the release date of a model or a guarantee that the information remains current. Search excerpts and inaccessible pages must not be presented as fully read evidence; the catalog marks those evidence limitations.
3. Compare the same thing
Identify the model or version, product surface, platform, and access route behind a claim. Do not transfer a feature from a model to every application that hosts it, or transfer a consumer-plan benefit to an API. Prices must retain their currency, metering unit, scope, and important conditions. A monthly subscription, a credit, a megapixel, and an output token are not interchangeable prices.
4. Keep unknowns and disagreements
An empty field means not established in the reviewed material, not unsupported or free. If official pages conflict, preserve the conflict and explain the relevant scope rather than silently choosing the more attractive claim. Deprecated endpoints and historical prices must remain clearly historical; their presence in the catalog is not a recommendation to adopt them.
5. Make recommendations proportional to evidence
A documented SVG export can justify shortlisting a tool for a vector workflow. It cannot establish that its drawings are the best. Workflow suggestions are editorial judgments based on stated criteria. We do not invent quality scores, speed measurements, popularity numbers, conversion rates, or universal winners from specification sheets.
6. Make comparisons reproducible
When comparing workflows, record the date, access plan, model/version, inputs, settings, selection rules, and evaluation criteria. Include correction time and the requirements of the finished deliverable so readers can apply the comparison to their own work.
7. Disclose AI assistance
AI assistance is part of the current research organization, drafting, and site-building process. It can introduce errors, omit qualifications, or misread a page. Source links and consistency checks help readers trace claims to the underlying documentation. No named human expert or staffed editorial team is implied by the brand byline.
8. Keep commercial influence visible
There are currently no affiliate links, paid placements, ads, or payment features. If any are introduced, disclose them in the affected content, including relevant relationships with a covered product’s operator. A sponsored inclusion must not be disguised as an independent test result. Payment is not evidence of quality.
9. Update and correct transparently
Review volatile availability, pricing, model versions, and terms before relying on an article for a new decision. This site does not promise continuous monitoring or a fixed refresh schedule. When a material correction is made, identify the changed claim, correct its source and scope, and add a dated correction note rather than merely changing the page date. Cosmetic edits should not imply a fresh factual review.
10. Make a correction actionable
A useful correction identifies the article, exact claim and supporting documentation. We keep the claim, its scope and its source together when revising coverage. Product terms and availability should be checked at the source before a purchasing or publishing decision.
11. Usage statistics and your choice
On studioverdict.app, optional Google Analytics helps us understand which guides readers use. Google’s script loads only after you choose Allow analytics. You can refuse without affecting access, or change your choice using Analytics preferences at the bottom of the page. If allowed, Analytics uses cookies and receives page usage and device information. We send page addresses without query strings or fragments and reduce referral addresses to their origin. Advertising features and Google Signals are disabled; we do not send a User-ID. A preference is stored in your browser to remember your choice. Private previews do not include Analytics. Google explains its handling of site data at https://policies.google.com/technologies/partner-sites.