3D Chess Media, LLC wants the information published across The 3D Chess Media Portfolio to be accurate, transparent, and capable of being corrected when material errors are identified. Our properties publish several different types of information, including:
- Original research
- AI recommendation data
- Citation data
- Industry benchmarks
- Company information
- Product and service information
- Comparative analysis
- Reviews
- Editorial commentary
- Commercial information
Different types of errors require different responses. This policy explains the general approach 3D Chess Media takes when a potential error is reported or discovered. Individual portfolio properties may maintain additional correction procedures appropriate to their subject matter.
Our Basic Principle
If we identify a material factual error, we want to correct it. That remains true whether the error benefits or harms: A client, An affiliate partner, An advertiser, A competitor, A sponsor, A 3D Chess Media business, and A company with no commercial relationship with us. Corrections should be based on the evidence. They should not depend on whether the affected company pays us.
What We Consider a Correction
A correction generally involves information that was inaccurate when published or became materially inaccurate because of a processing or reporting error. Examples may include:
- Incorrect company name
- Incorrect product name
- Incorrect price
- Incorrect executive title
- Incorrect ownership information
- Incorrect formation date
- Incorrect professional credential
- Incorrect calculation
- Duplicate data
- Missing data caused by processing failure
- Incorrect company matching
- Incorrect citation attribution
- Incorrect research date
- Incorrect description of a methodology
- Transcription error
- Data normalization error
When reliable evidence establishes that material information is incorrect, we may update the publication.
A Correction Is Different From an Update
Some information was accurate when published but later changed. That is usually an update, not a correction. Examples include:
- A company changes its pricing
- A product is discontinued
- An executive changes roles
- A company is acquired
- A website changes ownership
- An insurance product changes
- A software platform adds features
- An AI model changes its recommendations
The earlier information may have accurately reflected the conditions at the time it was collected. Where appropriate, the page may be updated to reflect the newer information.
A Correction Is Different From a Disagreement
A company may disagree with:
- Its ranking
- An AI recommendation result
- A comparison
- An editorial conclusion
- A research interpretation
- A reviewer's opinion
Disagreement does not automatically mean the publication contains a factual error. For example, if the underlying research accurately records that: 2 of 7 AI systems recommended Company A, Company A may believe it deserves to be recommended by all seven systems. That does not make: 2 of 7, a factual error if that was the observed result. We distinguish between: "The data was recorded incorrectly.", and "We do not like what the data says.", The first may require a correction. The second may not.
AI Responses Are Historical Observations
Some 3D Chess Media research records responses produced by external AI systems. Those responses can change. If ChatGPT recommended Company A on the date of collection and recommends Company B several weeks later, the earlier result was not necessarily incorrect. It may simply represent an earlier observation. Where practical, AI research should identify the collection period so readers understand when the measurement occurred.
Research Corrections
Research datasets can contain errors. Potential issues may involve:
- Prompt execution
- API failures
- Response collection
- Parsing
- Classification
- Entity matching
- Recommendation detection
- Citation extraction
- Ranking position
- Duplicate records
- Platform identification
- Data aggregation
- Calculations
If an error materially affects a published result, we may:
- Correct the underlying data.
- Recalculate affected metrics.
- Update the publication.
- Update charts or tables.
- Revise the conclusion if necessary.
- Note the correction when the change is material.
The appropriate response depends on the significance of the error.
Entity-Matching Errors
AI systems may refer to companies using different names. For example: ABC Software, ABC Technologies.
May refer to the same organization. Research sometimes requires entity normalization. A correction may be necessary if:
- Two separate companies were incorrectly combined
- One company was incorrectly split into multiple entities
- A product was attributed to the wrong parent company
- A similarly named company received another company's recommendation credit
- A citation was associated with the wrong organization
Entity corrections can materially change recommendation or citation totals.
Calculation Errors
If the underlying observations are correct but a mathematical calculation is wrong, we should correct the calculation. Examples may include:
- Incorrect percentage
- Incorrect average
- Incorrect ranking
- Incorrect total
- Incorrect share calculation
- Incorrect denominator
- Incorrect weighted score
- Incorrect month-over-month change
Where a corrected calculation changes the substantive conclusion, the accompanying text should also be reviewed.
Citation Corrections
Citation research may require corrections when:
- A URL is assigned to the wrong domain
- A citation is counted twice
- A source is missed
- A citation is attributed to the wrong response
- A redirect causes incorrect normalization
- Multiple URLs are incorrectly treated as one source
Because citation data can be aggregated across many responses, even relatively small classification errors can affect larger metrics.
Methodology Clarifications
Sometimes the data is correct but the methodology explanation is unclear. In those situations, we may update the publication to clarify:
- What was measured
- Which platforms were included
- How prompts were selected
- How rankings were calculated
- How citations were counted
- How companies were matched
- What dates were covered
- What limitations apply
A methodology clarification does not necessarily mean the underlying research result changed.
Material vs. Minor Corrections
Not every typo requires a formal correction notice. Minor changes may include: Spelling, Grammar, Punctuation, Formatting, Broken links, and Minor wording improvements. These may be corrected without a public correction notice when they do not change the substance. Material corrections are different. A correction may be considered material when it changes:
- A ranking
- A major factual claim
- A research conclusion
- A company identity
- A score
- A recommendation count
- A significant statistic
- A materially important disclosure
Material corrections may warrant an explanatory note.
Correction Notes
Where appropriate, a page may include a notice such as: Correction: An earlier version of this report incorrectly attributed one AI recommendation to Company A. The underlying dataset has been corrected and the recommendation totals recalculated. The level of detail should be proportional to the significance of the change.
We Do Not Require Payment for Corrections
A company does not need to: Become a client, Purchase advertising, Join an affiliate program, Sponsor research, Pay a correction fee, and Enter into a partnership. to report a potential factual error. Correction requests are evaluated based on the evidence provided.
Commercial Relationships Do Not Prevent Corrections
Commercial partners can be corrected. Clients can be corrected. Related companies can be corrected. 3D Chess Media properties can be corrected. A financial relationship does not make inaccurate information accurate.
Commercial Pressure Does Not Create a Correction
The reverse is also true. A company cannot purchase a correction simply because it dislikes an accurate result. We do not consider the following, by themselves, sufficient grounds for changing research:
- Threatening to end an affiliate relationship
- Refusing to advertise
- Threatening to stop purchasing services
- Offering sponsorship
- Requesting a better ranking
- Disagreeing with an observed AI recommendation
Commercial pressure is not evidence.
Related Companies
Because 3D Chess Media operates multiple businesses, one portfolio property may occasionally publish information concerning another. For example: LLM Authority Index may study AI search platforms or agencies., and CiteWorks Studio may appear in research concerning AI search services.. If a related company is incorrectly described or incorrectly measured, the correction should follow the same evidentiary standards applied to outside companies. Ownership should not provide immunity from correction.
Company-Supplied Corrections
Companies are often the best primary source for facts about themselves. We welcome reliable corrections concerning information such as:
- Company name
- Ownership
- Product availability
- Pricing
- Service areas
- Leadership
- Product specifications
- Official policies
Where possible, supporting evidence should come from authoritative sources such as:
- Official company pages
- Regulatory filings
- Government records
- Official documentation
- Public announcements
- Direct documentation supplied by an authorized representative
Company Claims Are Evaluated Separately
A company may provide factual information about itself. That does not automatically establish broader promotional claims as objective facts. For example: Our annual plan costs $499, may be directly verifiable. But: We are the most trusted company in America, may require independent evidence. We may correct factual information without adopting unsupported marketing claims.
Evidence for a Correction Request
A useful correction request should identify:
- Website
- Page URL
- Specific statement or data point
- Why it is believed to be incorrect
- Correct information
- Supporting source
- Contact information
The more specific the request, the easier it is to review.
Anonymous Correction Requests
We may review anonymous correction requests. However, we may need additional information to verify certain claims. For example, an individual claiming to represent a company may be asked to provide evidence of that relationship before we rely on non-public information.
Confidential Information
Do not send highly sensitive or confidential information through a general correction request unless it is necessary. If private documentation is required to verify a material correction, contact us first so an appropriate method of communication can be determined.
Legal Demands
A legal demand is not automatically proof that content is inaccurate. If we receive a legal complaint concerning published information, we may review:
- The statement
- The evidence
- The source
- The research record
- Applicable law
- The requested change
Where appropriate, we may consult legal counsel. A legitimate factual error should be corrected regardless of whether a legal threat was made. An accurate statement does not automatically become inaccurate because it is challenged.
Updates to Rankings
Rankings may change when:
- New data is collected
- New AI responses are measured
- A new platform is added
- A platform is removed
- A company becomes eligible
- A company becomes ineligible
- Methodology changes
- Product availability changes
A new ranking does not necessarily mean the old ranking was erroneous. It may represent a newer research period.
Versioning
Some research properties may maintain: Publication dates, Last-updated dates, Dataset versions, Research periods, and Snapshot identifiers. Versioning helps distinguish: a corrected historical dataset, from: a newly collected dataset. This distinction becomes increasingly important as AI search research develops over time.
Historical Research
We may preserve older research for historical comparison. An older report can remain useful even when current conditions have changed. Where practical, historical content should make clear that it reflects an earlier research period. We do not believe every historical study should be rewritten to reflect today's results. That would eliminate the historical record.
Retractions
In rare cases, a publication may contain problems significant enough that correction is insufficient. Potential examples might include:
- Fundamentally corrupted data
- Invalid research collection
- Material fabrication
- Serious methodological failure
- Content that cannot reasonably be repaired
In such circumstances, we may remove or retract the publication. A retraction is more serious than a normal correction and should be used accordingly.
Corrections and AI Training
We cannot control how quickly independent search engines, AI systems, caches, archives, or third-party services recognize a correction. Even after we update our page, an older version may continue to appear temporarily elsewhere. Our responsibility is to correct the information under our control.
How Quickly We Review Corrections
The time required depends on the issue. A simple factual correction may be relatively straightforward. A complex research dispute may require review of:
- Raw responses
- Prompt logs
- Data-processing records
- Calculation logic
- Historical versions
- Supporting sources
We prefer to investigate carefully rather than make an unsupported change quickly.
Requesting a Correction
To request review of a potential error, please include:
- Website: The 3D Chess Media property involved.
- Page: The full URL.
- Issue: The exact statement, number, ranking, citation, or data point you believe is incorrect.
- Correction: What you believe the accurate information should be.
- Evidence: Links or documentation supporting the correction.
- Contact: Your name, organization, and contact information where appropriate.
Contact for Corrections
3D Chess Media, LLC
Email Valerie@3DChessMedia.com
Telephone 949-309-0024
For faster review, include Correction Request and the relevant website name in the subject line.
Our Corrections Standard
Our standard is straightforward:
- Correct material factual errors when reliable evidence shows they are wrong.
- Do not rewrite accurate research because a company dislikes the outcome.
- Distinguish historical observations from current conditions.
- Preserve the underlying research record whenever possible.
- Explain material corrections when transparency requires it.
Related Policies
Portfolio & Ownership Disclosure