Corrections & Updates

AI Marketing Consensus Index publishes research in a market that changes quickly. AI platforms change. Models change. Marketing companies change. Software capabilities change. Pricing changes. Agencies expand or discontinue services. AI recommendation patterns change. For that reason, our correction and update policy distinguishes between three different events:

A Correction

We got something wrong.

An Update

The underlying company, product, or market changed after publication.

A Research Rerun

We asked the AI platforms the research question again at a later date and received a new set of responses. These events should not be treated as interchangeable.

Our Core Principle

Our corrections policy can be summarized as:

Correct errors. Record changes. Preserve historical research.

We want current articles to be accurate. We also want our historical AI recommendation data to remain useful. That means we should correct genuine mistakes without rewriting the past simply because the market has changed.

What Is a Correction?

A correction is appropriate when information published by AI Marketing Consensus Index was materially incorrect at the time it was published. Examples may include:

• Incorrect AI recommendation count
• Incorrect recommendation position
• Incorrect recommendation coverage percentage
• Duplicate recommendation
• Omitted recommendation
• Incorrect company normalization
• Incorrect company identity
• Incorrect ranking calculation
• Wrong Consensus Score
• Incorrect product capability
• Incorrect pricing information
• Incorrect company description
• Incorrect relationship disclosure
• Incorrect research date
• Incorrect reviewer attribution
• Incorrect citation or source attribution

When a material error is confirmed, we should correct it.

What Is an Update?

An update occurs when previously accurate information later changes. For example: A software platform may:

• Add ChatGPT tracking
• Remove a supported AI platform
• Introduce citation monitoring
• Change its pricing
• Launch a new product
• Discontinue a feature
• Change its target market
• Change its agency offering

An agency may:

• Add a new service
• Change positioning
• Merge with another company
• Stop offering GEO services
• Launch an AI visibility platform

Those changes do not necessarily mean our earlier reporting was wrong. They mean: the company changed.

What Is a Research Rerun?

A research rerun occurs when we conduct the AI recommendation study again. For example:

January Research

Company A — recommended by 6 of 7 platforms

April Research

Company A — recommended by 5 of 7

July Research

Company A — recommended by 3 of 7 Those are three different research snapshots. The January result does not become incorrect simply because the July result differs.

New AI Results Do Not Rewrite Old AI Results

AI Marketing Consensus Index is designed to measure recommendation behavior over time. Therefore, when research is rerun: do not overwrite the original response set. Preserve where practical:

• Original raw responses
• Original recommendation counts
• Original recommendation positions
• Original ranking
• Original research date
• Original methodology version

Then create a new research snapshot.

Historical Movement Is Part of the Dataset

Changes in AI recommendations can be valuable research. Over time, we may be able to measure:

• Which GEO agencies gain recommendation share
• Which AI visibility platforms lose recommendation share
• Which companies begin appearing in ChatGPT recommendations
• Which companies disappear
• Which tools become more prominent in enterprise use cases
• Which brands gain citation visibility
• Which competitors gain or lose category leadership

Deleting previous results would destroy that information.

Research Dates Must Remain Meaningful

Every study should preserve its actual: Research Date That date should represent when the underlying AI responses were collected. Updating a company feature does not change the original research date.

Separate Date Fields

Where practical, the site should maintain separate fields for:

researchDate

When the AI research was performed.

datePublished

When the page was originally published.

companyDetailsVerifiedDate

When material company, platform, or service information was last checked.

dateModified

When meaningful page content was changed.

reviewedDate

When Mark B. Huntley, J.D. completed applicable editorial review. These dates should not automatically update together.

Do Not Artificially Refresh Articles

Minor changes should not make old research appear new. Examples that should not automatically reset the primary research date include:

• Fixing punctuation
• Changing formatting
• Updating internal links
• Replacing an affiliate URL
• Changing a CTA
• Adjusting page design

We should not create artificial freshness by changing dates without meaningful new research or editorial work.

Research Updates and Company Updates Are Separate

Suppose the AI study was conducted in January. In March, Company A adds a new feature. We may update the company information and state: Company details verified March Not yet supplied. The underlying recommendation data still belongs to: January research. This distinction should remain visible.

AI Errors Are Not Automatically Our Errors

An AI platform may make a factual mistake. For example: “Company A monitors seven AI platforms.” Current company documentation may show that it monitors four. If we accurately recorded what the AI platform said, the AI response itself is not a data-processing error by AI Marketing Consensus Index. The correct treatment is: Preserve the AI recommendation. Then separately explain that:

current verification does not support part of the AI's reasoning.

Do Not Rewrite an AI Vote Because Its Reasoning Was Wrong

Suppose six AI systems recommend Company A. Three of them cite a capability that current verification does not support. The historical result remains: 6 of 7 AI platforms recommended Company A. We should not change it to: 3 of 7 simply because three recommendations relied on inaccurate reasoning. The recommendation occurred. The factual problem should be disclosed separately.

Verification Can Change Our Editorial Analysis

Although verification does not alter the underlying vote count, it can affect our interpretation. For example: Company A received the highest AI recommendation coverage in this study. However, several AI platforms attributed citation-tracking capabilities to the product that we could not verify in the company's current documentation. That is both:

• faithful to the research
• useful to the reader

Recommendation Extraction Errors Should Be Corrected

If our system incorrectly classifies a response, that is our error. Examples: An AI response says: “I would not recommend Company A.” but our system counts that as a positive recommendation. Or: “Company B is worth considering.” but our extraction system fails to count it. Those errors should be corrected. Affected calculations should then be recalculated.

Entity Normalization Errors Should Be Corrected

AI systems may use:

• Brand names
• Product names
• Parent company names
• Abbreviations
• Alternate spellings

If we incorrectly merge two different companies or incorrectly split one company into multiple entities, we should correct the normalization. That may affect:

• Recommendation count
• Ranking
• Category totals
• Company pages
• Historical analysis

Corrections Should Propagate

Whenever practical, structured research data should be centralized. If a correction changes:

• Recommendation count
• Company identity
• Average recommendation position
• Consensus Score

the correction should propagate to affected:

• Study pages
• Category rankings
• Company pages
• Homepage rankings
• Research summaries
• Related datasets

We should avoid correcting one page while leaving contradictory data elsewhere.

Material Corrections Should Be Disclosed

For meaningful corrections, we may add a note such as: Correction — Not yet supplied: We previously reported that Company A was recommended by 5 of 7 AI platforms. A review of the underlying responses found that one comparison reference was incorrectly classified as a recommendation. The correct result is 4 of 7. The ranking and related calculations have been updated. A correction note should explain:

• What was wrong
• What changed
• When it was corrected

without unnecessary defensiveness.

Minor Corrections May Not Require a Public Note

Minor changes such as:

• Typographical errors
• Broken links
• Formatting problems
• Minor grammatical fixes

may be corrected without a formal correction notice when they do not materially change the meaning of the content.

Company Changes Are Usually Updates, Not Corrections

Suppose an AI visibility platform charged: $199 per month when we verified it. Later, the company increases its price to: $299 per month. Updating the price is an: update not: a correction of the earlier article. The original price may have been accurate when published.

Product Launches Are Updates

If a company launches:

• A new AI citation feature
• A new agency service
• A new enterprise plan
• A new AI platform integration

we may update the current company information. That does not require changing historical AI recommendation counts.

Product Discontinuations Are Updates

Likewise, if a company:

• Removes a feature
• Discontinues a product
• Leaves a market
• Stops accepting new customers

we should update the current information when material. Historical research should remain preserved.

Company Rebrands and Acquisitions

If a company:

• Rebrands
• Changes its name
• Is acquired
• Merges with another company

we should preserve historical entity continuity where possible. The site should distinguish between: what the company was called when the research occurred and: its current identity.

AI Platforms Themselves Change

The systems we research also change. A platform may:

• Change models
• Add web search
• Remove search access
• Change recommendation behavior
• Change citation behavior
• Change interfaces

Those changes are one reason historical research should be dated and versioned.

Platform Changes Do Not Retroactively Alter Research

If an AI platform changes substantially after a study: the original result remains a valid record of what that platform produced at the time. We should not rerun historical calculations as though the new platform configuration existed in the past.

Changes to the Platform Set

Our standard research universe may evolve. If a future methodology uses a different platform set:

• Historical studies should retain their original denominator
• Historical recommendation rates should not be recalculated using platforms that were not part of the study
• New research should use the current methodology version

Methodology Changes Are Not Corrections

Suppose we later improve:

• Recommendation extraction
• Tie-breaking
• Category aggregation
• Qualification thresholds

That does not automatically mean every older study was incorrect. The appropriate response is: methodology versioning.

Methodology Versions

Where practical, each research snapshot should be associated with a methodology version. Example: Methodology 1.0 Methodology 1.1 If a major methodology change materially affects comparability, we should disclose that when comparing periods.

Do Not Retroactively Optimize Historical Rankings

We should not rewrite old rankings merely because a newer methodology would produce a more attractive result. Historical studies should remain tied to the methodology used when they were conducted.

Related Companies Receive the Same Correction Standard

AI Marketing Consensus Index has disclosed relationships with: LLM Authority Index and: CiteWorks Studio. Those companies may appear in research published on this site. They are subject to the same correction principles as unrelated competitors.

LLM Authority Index Cannot Privately Rewrite Its Results

LLM Authority Index provides AI research data and measurement infrastructure supporting AI Marketing Consensus Index. That relationship does not allow it to:

• Increase its own recommendation count
• Remove a competitor
• Reclassify an unfavorable AI response without evidence
• Delete historical results
• Change ranking calculations for commercial reasons

If LLM Authority Index identifies a legitimate error, the correction should be handled through the same evidence-based process applied to other companies.

CiteWorks Studio Cannot Rewrite Its Results

CiteWorks Studio provides AI search strategy and subject-matter support to AI Marketing Consensus Index. That relationship does not permit CiteWorks Studio to:

• Change its ranking
• Add recommendations
• Remove competitors
• Delete unfavorable results
• Turn an editorial disagreement into a correction

A correction requires evidence.

Related Companies Can Request Corrections

LLM Authority Index and CiteWorks Studio may identify:

• Incorrect feature descriptions
• Wrong pricing
• Incorrect entity mapping
• Data-processing problems

Those requests should be reviewed. But their related status does not create a lower correction threshold.

Competitors Have the Same Right to Request Corrections

A competing company does not need to:

• Advertise with us
• Become an affiliate
• Purchase data
• Work with CiteWorks Studio
• Work with LLM Authority Index

to request a correction. Corrections should depend on evidence, not commercial relationships.

Commercial Relationships Do Not Determine Corrections

A company cannot purchase:

• A correction
• Removal of an accurate ranking
• Competitor removal
• Historical deletion
• Favorable data treatment

Likewise, we should not refuse to correct an error because the affected company is not commercially valuable.

Paying Does Not Make a Complaint Valid

A commercial partner may dislike:

• Its ranking
• The companies that outranked it
• An unfavorable conclusion
• Weak recommendation coverage

That does not automatically establish an error. Corrections require evidence that our published data or factual reporting is actually wrong.

Disagreement Is Not Necessarily an Error

A company may say: “We believe our platform is better than Company A.” That is an opinion. If the research shows Company A received more AI recommendations, the ranking remains based on the research. Likewise: “We should have ranked higher.” is not itself evidence of a processing error.

Mark B. Huntley's Role in Corrections

Mark B. Huntley, J.D. serves as AI Search & Visibility Research Reviewer where indicated. Mark may help evaluate issues involving:

• AI search terminology
• GEO/AEO terminology
• Recommendation interpretation
• Citation interpretation
• Research conclusions
• Company-category fit
• Methodological context

His role does not include changing structured research data merely because he disagrees with the result.

Mark's Related Interests Require Additional Care

Mark has financial and operational interests associated with:

• LLM Authority Index
• CiteWorks Studio

When a correction materially affects one of those companies, the supporting evidence should be documented clearly. The correction should be traceable to:

• Raw research
• Calculation logic
• Current company documentation
• Other appropriate evidence

rather than editorial preference.

Reviewer Corrections and Data Corrections Are Different

Mark may identify an editorial issue such as: “This paragraph incorrectly describes citation visibility as recommendation visibility.” That can be corrected editorially. A change to: 5 of 7 recommendations requires evidence from the underlying research data.

Reviewer Dates Should Not Be Automatically Refreshed

If Mark previously reviewed an article, a minor product update should not automatically create a new: Reviewed by Mark B. Huntley — Not yet supplied date. A new reviewer date should mean meaningful review occurred.

Significant Updates May Require Re-Review

Certain updates may materially affect the interpretation of an article. Examples:

• Major product change
• Major company repositioning
• Material methodology change
• Substantial rewrite
• New research cycle

Those articles may be flagged: Re-Review Required before displaying a new reviewer date.

Suggested Update Workflow

For current company information: Change detected

Verify against reliable source

Update structured company data

Propagate affected fields

Update companyDetailsVerifiedDate

Update article if material

Request human re-review if interpretation materially changed

Suggested Correction Workflow

For a potential research error: Potential error identified

Open underlying raw response/data

Determine whether an actual error occurred

Correct structured record

Recalculate dependent metrics

Propagate changes

Add material correction note if warranted

Preserve correction history

Suggested Research Rerun Workflow

For a new research cycle: Use current approved research prompt

Run across current research-platform set

Create new raw response set

Extract new recommendations

Calculate new ranking

Verify material company information

Perform applicable editorial review

Publish as new research snapshot

Preserve previous snapshot

Company Requests for Updates

Companies may notify us that:

• Pricing changed
• Features changed
• Product positioning changed
• Supported AI platforms changed
• A service was discontinued
• A new service launched

We may evaluate the information and update the current factual layer where appropriate. Providing updated information does not entitle a company to changes in historical AI recommendation data.

Supporting Evidence

Correction requests are more useful when they include evidence such as:

• Page URL
• Description of the alleged error
• Correct information
• Supporting company documentation
• Relevant product page
• Public pricing page
• Technical documentation
• Other appropriate source

Recommended Correction Form Fields

The Contact page should support a: Correction option. When selected, request:

• Name
• Email
• Company
• Page URL
• Information believed to be incorrect
• Proposed correction
• Supporting Source URL
• Additional context

Corrections Should Not Depend on Company Participation

A company does not need to cooperate with our research for us to correct an obvious factual error. Likewise, a lack of company response does not prevent us from using reliable public evidence.

We May Decline Unsupported Correction Requests

We may decline to make a requested change when:

• The evidence does not support it
• The request concerns opinion rather than fact
• The request attempts to change accurate historical research
• The company simply disagrees with an AI recommendation
• The request asks us to suppress a competitor
• The request asks us to hide an accurate commercial disclosure

Related Business Disclosures Are Not Optional Corrections

CiteWorks Studio or LLM Authority Index should not request removal of accurate relationship disclosures simply because the disclosures are commercially inconvenient. If the relationship exists, the disclosure should remain. If the relationship changes, the disclosure should be updated accurately.

Ownership or Relationship Changes

If the relationship among:

• AI Marketing Consensus Index
• Mark B. Huntley, J.D.
• LLM Authority Index
• CiteWorks Studio

materially changes, the relevant disclosure pages and article-level notices should be updated. Historical articles may preserve the relationship disclosure that was accurate at the time where relevant.

Changes to Affiliate Relationships

Beginning or ending an affiliate relationship does not require changing:

• Historical ranking
• AI recommendation count
• Consensus Score

It may require changing:

• Affiliate links
• Commercial disclosures
• CTA destination

Changes to Advertising Relationships

Beginning or ending advertising does not change the research result. A company should not gain rank when advertising begins or lose rank when advertising ends.

Data Licensing Relationships Do Not Change Research

A company purchasing:

• Research data
• Benchmarking
• Historical data
• Custom reports

does not gain the right to change public research.

Current Information and Historical Research Can Coexist

A study can simultaneously say: Research conducted January 12, 2027: Company A was recommended by 6 of 7 AI platforms. and: Company information verified April 8, 2027: Company A has since changed its product offering. Both statements can be true.

Correction History

Where technically practical, maintain internal records showing:

• Correction date
• Field changed
• Old value
• New value
• Reason
• Evidence
• Person or process approving correction

This is especially important for structured ranking data.

Public Correction Notes

Not every internal change needs a public log. But material changes affecting:

• Ranking
• Recommendation coverage
• Company identity
• Major factual conclusion

should generally receive an understandable public note.

We Do Not Delete Embarrassing Research

Historical research should not be removed simply because:

• A related company performed poorly
• A commercial partner lost rank
• A competitor performed well
• A newer result is more favorable

If the historical research was legitimately conducted and accurately reported, it remains part of the research record.

We May Remove Content for Legitimate Reasons

Content may be removed or substantially changed for reasons such as:

• Serious legal issue
• Privacy concern
• Security issue
• Duplicate publication
• Fundamentally invalid research
• Technical corruption
• Other exceptional circumstances

Where removal materially affects historical research, we should preserve an internal record of why it occurred.

Contact Us About an Error

If you believe something on AI Marketing Consensus Index is incorrect, please contact us. Useful information includes:

• Page URL
• Exact statement or data point
• Why you believe it is incorrect
• Current supporting source

Submit a Correction →

Our Corrections Standard in One Sentence

AI Marketing Consensus Index corrects genuine errors, updates current company information when the market changes, treats new AI research as a new historical snapshot, and does not rewrite accurate research because a new result or commercial interest is more convenient.

Related Business Transparency

AI Marketing Consensus Index operates with disclosed relationships relevant to this policy. LLM Authority Index provides AI research data and measurement infrastructure. CiteWorks Studio provides AI search strategy and subject-matter support. Mark B. Huntley, J.D. serves as AI Search & Visibility Research Reviewer where indicated and has financial and operational interests associated with these businesses.

These relationships do not provide:

• Special correction rights
• Ranking control
• Competitor-removal authority
• Historical deletion rights
• Additional recommendation credit

Corrections involving related companies should be supported by the same underlying evidence required for unrelated companies.

Read Related Business & Conflict of Interest Disclosure →

Related Policies

How We Rank →

Research Methodology →

Editorial Standards →

Editorial Independence →

Human Review Policy →

Related Business & Conflict of Interest Disclosure →

How LLM Authority Index and CiteWorks Studio Contribute →

Data & Research Limitations →

Affiliate Disclosure →

Advertising & Sponsorship Disclosure →

Contact Us →

Related-business relationships do not influence correction standards, AI recommendation data, or ranking calculations.

Related business disclosure · Research methodology