Skip to content
TruthVouch Docs

Narrative Tracking

A narrative is a consistent story that AI systems tell about your brand. It’s not a single fact — it’s a cluster of related claims that, together, paint a picture of who you are.

Understanding Narratives

Example Narrative: “Enterprise AI Safety Leader”

This narrative might be composed of these individual claims:

  • “They focus on AI safety”
  • “They serve enterprise customers”
  • “They were founded by Stanford AI researchers”
  • “They’re SOC 2 Type II certified”
  • “They have published research on hallucination detection”

When ChatGPT, Claude, and Gemini all mention 3+ of these claims together, Brand Intelligence clusters them into a narrative.

Narratives vs. Individual Claims

Individual ClaimNarrative
”Founded in 2020""Founded by AI researchers with strong academic pedigree"
"Serves 500 customers""Growing rapidly, enterprise-focused"
"Offers API""Developer-friendly platform for AI integration"
"Open source library""Transparent, community-driven approach”

A narrative is bigger than any single claim — it’s the story AI tells when asked “What is this company?”

Narrative Dimensions

Each narrative tracked in Brand Intelligence has several dimensions:

1. Polarity

  • Positive: “An innovative leader in AI governance”
  • Neutral: “A B2B SaaS company focused on compliance”
  • Negative: “An expensive solution with limited integration”

2. Prevalence

How widespread is this narrative across AI engines?

Prevalence = (# of engines mentioning this narrative / Total engines monitored) × 100

Example:

  • ChatGPT: “Leader in AI safety” ✓
  • Claude: “Leader in AI safety” ✓
  • Gemini: “A compliance-focused AI tool” ✗
  • Perplexity: “Leader in AI governance” ✓

Prevalence = 75% (3 out of 4 engines mention it)

3. Trend

Is a narrative gaining or losing traction?

Trend = This week's prevalence - Last week's prevalence
  • Growing (+15%): More engines are repeating this narrative now
  • Stable (±5%): Consistent presence across engines
  • Declining (-10%): Fewer engines are mentioning it

Emerging narratives (new in the last month) are highlighted separately.

4. Strength

How confident are AI engines in this narrative?

Some narratives are stated as fact:

“They are enterprise security leaders.” (High strength)

Others are hedged:

“They seem to focus on compliance.” (Low strength)

Brand Intelligence measures strength by analyzing the certainty language and claim density.

Dashboard Narrative View

Navigate to Brand Intelligence → Dashboard → Narratives.

You’ll see:

Your Top 5 Narratives

Sorted by prevalence (most widespread first).

For each narrative:

  • Narrative title (auto-generated or custom)
  • Polarity badge (green=positive, gray=neutral, red=negative)
  • Prevalence bar (% of engines mentioning it)
  • Trend arrow (↑ growing, ↓ declining, → stable)
  • First detected (when this narrative appeared)

Detailed View

Click any narrative to see:

  1. Examples — Exact phrases from AI engine responses

    • “ChatGPT: ’…’”
    • “Claude: ’…’”
    • “Gemini: ’…’”

  2. Component claims — The individual facts that make up the narrative

    • ✓ Verified (matches truth nugget)
    • ✗ Contradicted (conflicts with truth nugget)
    • ? Unverified (not in your truth nuggets, but mentioned in AI responses)

  3. Trend chart — Week-over-week prevalence changes

  4. Recommended actions — How to strengthen or counter the narrative

Narrative Types

Positive Narratives

These work for you. AI systems are saying good things about your brand.

Examples:

  • “Leaders in AI security”
  • “Enterprise-grade compliance”
  • “Innovative approach to hallucination detection”
  • “Transparent pricing model”

Actions:

  1. Strengthen: Create content that reinforces these narratives
    • Blog posts highlighting your security features
    • Case studies with enterprise customers
    • Whitepapers on your AI innovation
  2. Amplify: Make sure positive claims are prominent on your website
  3. Monitor: Watch the trend line to ensure they don’t decline

Neutral Narratives

These are factual descriptions of your business. They’re neither positive nor negative.

Examples:

  • “B2B SaaS company”
  • “Serves mid-market and enterprise”
  • “Focused on AI governance”
  • “Founded in 2020”

Actions:

  1. Keep current: Ensure these facts remain accurate
  2. Clarify if needed: If a neutral narrative is misleading (e.g., “Startup” when you’re actually 5 years old and profitable), add context
  3. Contextualize: If a neutral narrative is outdated (e.g., “Early-stage company” but you’re now Series C), update your website

Negative Narratives

These could hurt your brand if inaccurate or if they’re preventing positive narratives from emerging.

Examples:

  • “Expensive solution”
  • “Limited integration options”
  • “Requires significant implementation effort”
  • “Small customer base” (if untrue)

Actions:

  1. Verify accuracy: Is this narrative based on fact or perception?
  2. If accurate but dated: Update your website to address the concern
    • “Expensive” → Show ROI/value case study
    • “Limited integrations” → Add new integrations, document them
  3. If inaccurate: Create content that corrects the record
    • “Limited integrations” → “Integrates with 40+ platforms”
  4. If opinion-based: Don’t fight it directly; address the underlying concern
    • “Expensive” → Focus on why you’re worth it (security, compliance, reliability)

Emerging vs. Persistent Narratives

Emerging Narratives

New narratives detected in the last month, not yet widespread.

Why they matter:

  • Emerging narratives can become persistent if they spread unchecked
  • Early intervention is cheaper than correcting entrenched narratives

Example timeline:

  • Week 1: One engine mentions “Complex setup process”
  • Week 2: Two engines mention it
  • Week 3: Three engines mention it
  • Week 4: Four engines mention it (becoming persistent)

Action: If an emerging negative narrative is inaccurate, address it immediately by:

  1. Improving website documentation or onboarding instructions
  2. Publishing a blog post titled “Getting Started: [Product] in 10 Minutes”
  3. Creating an FAQ addressing the concern

Persistent Narratives

Narratives that have been present for 3+ months and are mentioned by 50%+ of engines.

These are harder to change, but still possible.

Action:

  1. Identify the source: Where did this narrative originate? Often an old blog post or news article.
  2. Counter with newer sources: Publish content that supersedes the old narrative
  3. Use structural data: Mark-up your content so AI systems know what’s current vs. old
  4. Monitor implementation: Check back weekly; persistent narratives take 2-3 months to shift

Contamination Risk

A narrative becomes a contamination risk when:

  1. It’s false or outdated
  2. It’s mentioned by 3+ engines (already spreading)
  3. It’s persistent (been around 2+ months)

Example of contamination risk: Week 1: “Founded in 2018” mentioned by ChatGPT (sole source: old blog post) Week 3: Now mentioned by Claude and Gemini too (they’re copying ChatGPT’s training data) Week 8: Now mentioned by Perplexity (it appears in multiple web sources) Month 3: Five engines repeat “Founded in 2018”

By month 3, this false narrative is hard to change because it’s entrenched across multiple AI training data sources.

Prevention: Fix inaccuracies when they appear in just one engine (low contamination risk), not after they spread to five engines (high contamination risk).

Taking Action on Narratives

Positive Narrative: “Enterprise AI Safety Leader”

Current status: 75% of engines mention it, growing

Action:

  1. Make this narrative MORE prominent on your website
  2. Add case study: “How [Customer] Uses [Product] for AI Safety”
  3. Publish blog post: “Enterprise AI Safety: 5 Critical Controls”
  4. Ensure all product pages link to security/safety documentation

Expected result: Prevalence increases from 75% to 90%+

Negative Narrative: “Limited Integration Options”

Current status: 50% of engines mention it, stable

Check: Is this true?

  • If true → Add integrations, document them, publish blog post, update website
  • If false → Check where this comes from. Often an old comparison chart or demo site.

Action (if false):

  1. Create page: “[Product] Integrations: 40+ Connections”
  2. Publish blog: “Expanding Integrations: May 2024 Update”
  3. Update homepage to highlight integrations
  4. Check back in 2 weeks

Expected result: Prevalence drops from 50% to 25%+ over 3 months

Narrative Reporting

Your weekly digest includes:

  • Emerging narratives: New narratives detected
  • Narrative movement: Which narratives grew, declined, or stayed stable
  • Contamination risk alerts: When a narrative crosses from low-risk to high-risk spreading

You can customize narrative reporting in Brand Intelligence → Settings → Reports.

Common Questions

”Why do engines have different narratives about us?”

Different AI models have different training data, training dates, and response styles. GPT-4o and Claude have April 2024 cutoffs. Gemini has a late 2023 cutoff. So they might have different versions of your story.

”How do I change a narrative?”

You change narratives by changing your website and other authoritative sources. AI learns from those sources, so updating your website is the most direct way to shift narratives.

”Can a narrative be both positive and negative?”

Yes. “Startup” can be positive (innovative, agile) or negative (not proven, risky). The polarity depends on context.

”How long does it take to change a narrative?”

Emerging narratives can be corrected in 2-4 weeks. Persistent narratives take 2-3 months. It depends on how widely the narrative has spread through AI training data.

Next Steps