# AI answer accuracy auditor Human Guide

## What This Is For
Audits factual claims and editorial framing about a brand across AI answers, traces likely source-of-truth failures, and produces a prioritized correction ledger. It gives the agent a clearer input/output frame for AI answer accuracy auditor: what context to ask for, what decisions to make, and what usable artifact to return.

Use this as a human-readable version of the AI answer accuracy auditor agent skill. It is meant for marketers, operators, founders, and other non-coders who want the workflow without reading agent-specific implementation instructions.

## When To Use This
- Use this when you need a repeatable process for AI answer accuracy auditor.
- Use this when the task needs judgment, examples, constraints, or a clear output format rather than a one-off prompt.
- Use this when you want to hand an AI assistant enough context to produce a usable marketing artifact.

## When Not To Use This
- Do not use this when you only need a quick factual answer.
- Do not use this when the work depends on private data you cannot share with the assistant.
- Do not use this as a replacement for legal, compliance, financial, or medical review.

## What You Need Before Starting
- The goal or business outcome you want.
- The audience, customer segment, or market context.
- Any source material the assistant should respect, such as notes, briefs, examples, URLs, or brand guidance.
- Constraints such as tone, length, channel, deadline, region, or approval requirements.
- A clear definition of what a good final answer should look like.

## Step-By-Step Workflow
1. State the job clearly: "Use the AI answer accuracy auditor guide to help me with..."
2. Add context: audience, goal, offer, channel, source material, and constraints.
3. Ask the assistant to identify missing inputs before producing the final output.
4. Have the assistant follow the skill-specific guidance below.
5. Review the result against the final checklist and ask for revisions where needed.

## Skill-Specific Guidance
- Across 50,000 prompts, about 47% of response content was unsolicited editorial material such as comparisons, rankings, caveats, and recommendations.
- A brand can be visible while still being described inaccurately or framed poorly.
- Specific, current, verifiable first-party facts give answer engines stronger material than generic marketing claims.
- Profound recommends recurring claim audits, source-of-truth updates, and human-reviewed correction workflows.
- products or services in scope
- engines, markets, and languages
- captured AI answers or permission to research them
- current product documentation, pricing, policies, and approved claims
- competitors and high-risk topics
- Define the audit cohort by engine, market, language, prompt family, and date.
- Extract every decision-relevant claim, including unsolicited editorial claims.
- Label each claim: requested fact, comparison, recommendation, ranking, caveat, price, availability, compatibility, limitation, or sentiment.

## Decision Points And Nuance
The original skill emphasizes: Research basis, Goal, Intake, Workflow, Severity, Output.

Use these questions to steer the work:
- What is the intended audience or buyer?
- What source material must be preserved?
- What should the assistant optimize for: clarity, persuasion, accuracy, speed, creativity, or conversion?
- What examples represent the desired quality bar?
- What should the assistant avoid?

## Common Mistakes
- Never treat the brand's preferred wording as proof. Use official product material for first-party facts and reliable independent sources for external claims.
- Do not fabricate corrections, manipulate Wikipedia, astroturf communities, or pressure publishers to remove legitimate criticism.
- **High:** false comparisons, wrong limitations, or stale product facts in important prompt clusters.
- Never promise that correcting a source will change an AI answer. Report observed movement over time.

## Copy-And-Paste Prompt
```text
Use the AI answer accuracy auditor human guide.

My goal:
[Describe the business outcome]

Audience:
[Describe who this is for]

Context and source material:
[Paste notes, examples, links, or existing copy]

Constraints:
[Tone, length, channel, timeline, must-include items, must-avoid items]

Before producing the final output, ask me for any missing information that would materially improve the result.
```

## Final Checklist
- [ ] The output matches the original goal.
- [ ] The audience and context are reflected in the answer.
- [ ] Important constraints and source material were preserved.
- [ ] The assistant made the relevant decisions explicit.
- [ ] The final artifact is ready to use, review, or hand to the next person.

## Source
This guide was generated from the Profound skill entry for `ai-answer-accuracy-auditor`.

## Source Skill Notes
These notes preserve the nuance from the original skill. Use them as supporting reference when the workflow above feels too generic.

# AI Answer Accuracy Auditor

## Research basis

This skill operationalizes Profound's research, [The Parrot Problem: Why AI Search has a second dimension marketers can't ignore](https://www.tryprofound.com/blog/the-parrot-problem), published June 25, 2026.

Key takeaways from the study:

- Across 50,000 prompts, about 47% of response content was unsolicited editorial material such as comparisons, rankings, caveats, and recommendations.
- A brand can be visible while still being described inaccurately or framed poorly.
- Specific, current, verifiable first-party facts give answer engines stronger material than generic marketing claims.
- Profound recommends recurring claim audits, source-of-truth updates, and human-reviewed correction workflows.

Treat these findings as observational evidence, not universal model behavior. Record engine, model or surface, locale, prompt, and capture date for every answer.

## Goal

Find material claims about a brand, verify them against current evidence, diagnose where errors originate, and recommend the smallest legitimate correction.

## Intake

Ask only for missing inputs that block the audit:

- brand and canonical domain
- products or services in scope
- engines, markets, and languages
- captured AI answers or permission to research them
- current product documentation, pricing, policies, and approved claims
- competitors and high-risk topics

Never treat the brand's preferred wording as proof. Use official product material for first-party facts and reliable independent sources for external claims.

## Workflow

1. Define the audit cohort by engine, market, language, prompt family, and date.
2. Extract every decision-relevant claim, including unsolicited editorial claims.
3. Label each claim: requested fact, comparison, recommendation, ranking, caveat, price, availability, compatibility, limitation, or sentiment.
4. Verify the claim against the strongest current source available.
5. Classify the result:
   - accurate and current
   - accurate but incomplete
   - ambiguous
   - stale
   - false
   - subjective framing
   - unverifiable
6. Trace the likely source: owned page, documentation, feed, profile, third-party publisher, community page, or unknown.
7. Score severity using decision impact, exposure, confidence, and correction difficulty.
8. Fix the source of truth before proposing cosmetic rewrites.
9. Define a retest cohort and observation window.

Do not fabricate corrections, manipulate Wikipedia, astroturf communities, or pressure publishers to remove legitimate criticism.

## Severity

- **Critical:** safety, legal, eligibility, price, availability, or compatibility errors that can materially harm a decision.
- **High:** false comparisons, wrong limitations, or stale product facts in important prompt clusters.
- **Medium:** incomplete or misleading framing with plausible decision impact.
- **Low:** stylistic sentiment or immaterial wording.

## Output

Return:

1. Executive finding: accuracy rate, highest-risk narrative, and primary source-of-truth problem.
2. Claim ledger with engine, prompt, claim, type, verdict, evidence, likely source, severity, and owner.
3. Source correction plan with exact facts and canonical destinations.
4. Third-party correction plan limited to factual, evidence-backed outreach.
5. Retest plan using the same prompt cohort plus controlled variants.
6. Unknowns and limitations.

Never promise that correcting a source will change an AI answer. Report observed movement over time.