1. Philosophy of Accuracy

WhichAIPick strives for 100% accuracy in our data and analysis. However, we are analyzing a chaotic, rapidly evolving industry. Errors—whether mathematical, typographical, or interpretive—can occur. We believe that the measure of a publication's integrity is not whether it makes mistakes, but how transparently it corrects them.

We do not "stealth edit." If we change a material fact in a story, we note the change. History is not mutable. This aligns with our commitment to Data Transparency.

2. The Three Tiers of Errors

We categorize errors into three distinct tiers, each with a specific protocol for remediation.

Tier 1: Minor/Cosmetic (No Log Required)

Examples: Spelling mistakes, broken links, formatting issues, minor grammar syntax.

Protocol: We fix these immediately upon discovery. No "Correction" note is appended to the article, as these errors do not materially affect the reader's understanding of the subject matter.

Tier 2: Factual/Substantive (Correction Note Required)

Examples: Getting a price wrong ($29 instead of $39), misstating a feature (claiming a tool has an API when it doesn't), misspelling a founder's name.

Protocol: We correct the text in the article. We append a "Correction" footer at the bottom of the article explicitly stating what was changed and when.

Example Footnote: "Correction (Feb 19): An earlier version of this review incorrectly stated that Tool X offers SSO. They do not. We have updated the score to reflect this."

Tier 3: Fundamental/Critical (Retraction Required)

Examples: A methodology failure that invalidates the entire review, falsely accusing a vendor of data misuse, complete misunderstanding of the tool's purpose.

Protocol: This is the nuclear option. If a review is fundamentally flawed, we do not simply patch it. We may issue a full Retraction in the headline and leave the original text with a strikethrough or remove it entirely, replacing it with a transparency statement explaining the failure.

3. The "Vendor Right of Reply"

We are fair. If a vendor feels they have been unfairly characterized in a review, they have the right to challenge our findings. This is not a guarantee that we will change the score, but it is a guarantee that we will hear the case.

The Protocol for Vendors:

What We Do NOT Change: We do not change "Subjective" analysis. If our reviewer found the UI "cluttered and confusing," providing a screenshot showing that it is "actually complex and powerful" will not change our mind. That is an editorial opinion regarding UX.

4. What We Don't Do (Boundaries)

We Honor

  • Factual Evidence: If you prove we are wrong, we fix it.
  • Timestamps: We respect the version of the tool we tested.
  • Ombudsman Process: Escalating disputes to the board.

We Reject

  • Legal Threats: We do not respond to intimidation.
  • PR Pressure: "Can you change the tone?" earns a ban.
  • SEO Requests: "Can you change the anchor text?" No.

5. Community Reporting

Our readers are our best fact-checkers. We encourage our community to report errors. If you spot a discrepancy in pricing, features, or performance metrics, please use the "Report" link found in the footer of every page.

We maintain a "Hall of Fame" (internal) for users who help us catch significant errors, sometimes rewarding them with free access to our premium Academy Resources.


Related Resources

Version Notes:
v1.2 (2026-02-19): Final Lock. Added "Boundaries" matrix and Ombudsman contact.
v1.1 (2026-02-19): Formalized "Three Tiers" of error classification.
v1.0 (2024-02-01): Initial corrections protocol.