A useful claim has a visible source.
How we label evidence, disclose our commercial interests, publish research within its limits and correct factual errors.
Who is publishing this material
Readers should be able to see who makes a claim, what supports it and where the evidence stops.
BeautyGEO is a commercial generative engine optimization service provider for beauty brands and their supply chain. We are the publisher of this website and have a business interest in the services described here.
Research about our own visibility is first-party research. A comparison that includes BeautyGEO is self-affiliated. A company's own website can establish what that company publicly says; it does not independently verify every claim it makes. We distinguish those source types when drawing conclusions.
Label the kind of evidence
Official platform documentation
An attributable description of a provider's published mechanism or policy. Cite the exact page and review date. It does not prove that the mechanism has produced a result for our site.
Observed test result
What was captured from a defined run or report. State platform, surface, model where known, prompt, conditions, date and sample size. An excerpt supports only what the excerpt exposes.
Company-supplied fact
A statement from a company's website or supplied document. Name the source, its date and scope. Keep self-description distinct from independently verified evidence.
Analysis or recommendation
Our interpretation of evidence or a proposed action. Explain the reasoning and uncertainty; do not present a content tactic as a platform's undisclosed ranking rule.
Illustration or example
A diagram, staged visual, hypothetical calculation or demo. Label it clearly. Illustrative factory images are not evidence of our facilities, clients or production capabilities.
Unknown or not verified
Evidence is missing, inaccessible, contradictory or out of scope. Keep the limitation visible. Do not replace missing observations with a zero, a guessed value or a claim of universal absence.
Match the claim to what the source can support
| Claim | Evidence needed | Limit to keep visible |
|---|---|---|
| A crawler can access a page | Published bot controls plus the relevant request, response and verified request identity when claiming a real crawler visit. | An ordinary request with a bot-like user agent does not authenticate that bot or prove indexing. |
| An AI answer mentioned or cited a business | The dated answer, its test conditions, the correct entity and the specific source link where claimed. | A source list, an API sample or a supplied-URL read cannot silently become a consumer recommendation. |
| A supplier holds a certificate or offers a capability | An attributable document or issuer record tied to the correct entity, facility, scope and validity period. | A logo, an AI answer or a general industry standard does not verify that supplier's current status. |
| Our work improved results | Comparable observations with the change history, sample size and relevant limitations. | A correlation or before-and-after sample alone does not establish causation or a guaranteed future outcome. |
| A commercial result occurred | Authorized records establishing the relevant visit, qualified inquiry or verified revenue, with the attribution rule. | A test form, citation, lead estimate or proposal value is not a completed sale. |
Make comparisons and research limitations visible
Comparisons
- Disclose when BeautyGEO is a compared provider. State selection criteria and what the comparison covers.
- Link factual statements to the relevant company's own page or another attributable source. Treat advertised capabilities and prices as dated statements.
- Use consistent criteria. A capability not found on a page is not proof that a business lacks it.
- Do not imply endorsement, certification, a client relationship or independent ranking without evidence. Identify a relevant commercial or affiliate relationship when one exists.
Research and own-site tests
- Publish the scope: whose visibility, which prompts, platforms, dates and interface or API were tested.
- Show unsuccessful and unavailable states alongside the applicable results. Keep untested platforms marked as untested.
- Current public studies contain selected records, excerpts or aggregate summaries. They are not a complete public raw-response archive.
- A published method is a protocol, not evidence that a past study satisfied every requirement. Refer to the actual study record for coverage.
Read the measurement methodology for cohort definitions, counting rules and an empty observation template.
Keep permissions attached to the evidence
Client material
Confirm the agreed scope before publishing a named client, screenshot, metric, quote or supporting file. An internal analysis or a permission to review material does not automatically authorize public publication.
Private records
Exclude access tokens, account identifiers that reveal private access, personal contact details and confidential commercial material from public exports. Label redactions that materially limit interpretation.
Evidence availability
State whether a source is public, privately reviewable with permission or unavailable. Do not describe a restricted or missing file as an open dataset. Agree evidence delivery and retention in the engagement scope.
Correct the claim and preserve its context
Report the exact statement
Send the page URL, quoted statement, proposed correction and a supporting source to [email protected]. Include relevant dates, entity names or certificate scope so we can identify the issue.
Check the evidence
Review the original source and the proposed replacement. Where the evidence is incomplete, qualify the claim or mark it unresolved rather than treating an assertion as a verified correction.
Update the affected content
Correct or remove a substantiated error, and check associated summaries, structured data and downloads. Material changes should identify what changed and when; a new review date alone does not explain a changed result.
When sources conflict, retain the disputed scope and review date while the claim is being checked.
Use clear, supported content for people and machines
Visible copy, metadata, structured data and downloadable summaries should agree on the business identity and the facts. We do not use invented reviews, certifications, customer results or hidden instructions to make AI systems recommend us. AI-assisted drafting does not change the evidence standard; generated text still requires source and claim review.
Google's structured-data policies require markup to represent the visible content accurately. Bing's webmaster guidelines describe eligibility and quality practices without guaranteeing citations. The Organization and Service vocabularies help identify the publisher and service provider; they are not independent evidence of expertise or performance.
Neither an llms.txt file, a robots permission, a valid schema check nor a publication on this website guarantees discovery, indexing, a citation, a recommendation or revenue. Provider mechanisms are reviewed in the six-platform guides, with official documentation distinguished from our implementation advice.