Skip to content
TruthVouch Docs

Monitor What AI Says About Your Brand

ChatGPT, Gemini, Perplexity, and other AI engines are talking about your brand — but you have no idea what they’re saying. Outdated product info, wrong pricing, forgotten competitors, misattributed features. This guide shows you how to monitor what 9+ AI engines are saying about you in real-time and fix inaccuracies automatically.

Overview

Brand Intelligence monitors how 9+ AI engines represent your organization. Using your Truth Nuggets as ground truth, it continuously:

  • Sends cross-checks to AI engines asking factual questions about you
  • Compares responses against your verified facts
  • Detects hallucinations and inaccuracies
  • Alerts your team to issues
  • Deploys corrections automatically

The result is a Brand Trust Score (0-100) that measures how accurately AI knows your brand. With this score, you can:

  • Track brand accuracy trends over time
  • Prioritize which facts AI engines are getting wrong
  • Measure the impact of GEO optimization and content updates
  • Benchmark yourself against competitors

Prerequisites

  1. TruthVouch account with Professional tier or higher
  2. Truth Nuggets created for your organization (15-30 initial facts)
  3. Monitoring activated (happens automatically during onboarding)

If you haven’t set these up, start with Account Setup or For Marketing Leaders.

Step 1: Define What You Want Monitored

Your Truth Nuggets define what gets monitored. Make sure your most important brand facts are represented.

  1. Go to Knowledge Base → Truth Nuggets

  2. Review your existing Nuggets across categories:

    • Company facts: Founding date, headquarters, leadership, stage
    • Product facts: Names, launch dates, pricing, key features
    • Brand facts: Positioning, differentiators, awards, partnerships

  3. Add any critical facts TruthVouch missed:

    • “Founded in 2021” (if auto-extraction only found “founded in early 2020s”)
    • “Acquired by Acme Corp in March 2026” (recent news)
    • “Named a Gartner Magic Quadrant leader” (recent achievement)
    • “Pricing starts at $349/month” (if you just changed it)

  4. Prioritize facts:

    • Mark facts as Critical that would seriously impact your brand if misrepresented (CEO name, company status, core product claims)
    • Mark as Important facts that are commonly asked (pricing, feature list)
    • Mark as Standard facts that are nice-to-know (partner integrations, team size)

Critical facts will be cross-checked more frequently.

Step 2: Configure Cross-Check Schedules

Set how often TruthVouch queries AI engines about your facts.

  1. Go to Brand Intelligence → Cross-Check Settings

  2. For each AI engine, set the schedule:

    • ChatGPT (most used): Default to every 1-2 hours for critical facts, daily for others
    • Gemini (Google’s AI): Every 4-6 hours (slower to update)
    • Perplexity (AI search): Every 24 hours (good coverage but slower updates)
    • Claude (Anthropic): Every 24 hours
    • Bing Chat (Microsoft): Every 24 hours (mirrors web search freshness)

  3. For critical facts (CEO, founding date, core product), use:

    • ChatGPT: Every 1-2 hours
    • Other engines: Every 24 hours

  4. For product features and pricing, use:

    • ChatGPT: Every 24 hours
    • Others: Every 48-72 hours

Higher frequency = more API usage and slightly higher costs, but faster hallucination detection.

Step 3: Monitor Your Brand Dashboard

Watch your Brand Trust Score and cross-check results in real-time.

  1. Go to Brand Intelligence → Dashboard

  2. You’ll see:

    • Overall Brand Trust Score (0-100): Weighted accuracy across all engines
    • Score by engine: Individual scores for ChatGPT, Gemini, Perplexity, etc.
    • Recent cross-checks: Last 50 queries and their results
    • Top hallucinations: Which facts AI engines are getting wrong
    • Trend: Month-over-month improvement or decline

  3. Click on any hallucination to see:

    • What TruthVouch asked (the cross-check prompt)
    • What the AI engine responded
    • Your verified fact (from Truth Nuggets)
    • Confidence score (how certain is TruthVouch that this is a hallucination?)

Example cross-check:

  • Question: “What is the founding date of TruthVouch?”
  • ChatGPT response: “TruthVouch was founded in 2022”
  • Your fact: “Founded in 2021”
  • Confidence: 94%
  • Severity: High

Step 4: Set Up Alerts

Get notified immediately when hallucinations are detected.

  1. Go to Brand Intelligence → Alerts

  2. Configure by severity:

    • Critical: CEO changes, company status changes (acquired, shut down), founding date changes
    • High: Core product changes, major feature removal, significant pricing changes
    • Medium: Minor feature inaccuracies, secondary product claims
    • Low: Spelling issues, minor details

  3. Choose alert channels:

    • Email: To your team inbox
    • Slack: To a dedicated #ai-brand channel
    • SMS: Critical alerts to your phone
    • Webhooks: To custom systems

  4. Set quiet hours (optional):

    • No alerts between 6 PM and 9 AM
    • No weekend alerts
    • Custom schedules per severity level

Recommended: Critical alerts go to Marketing & Comms lead + CMO. High alerts go to team Slack channel. Medium and Low alerts digest to daily email.

Step 5: Review and Approve Corrections

When hallucinations are detected, TruthVouch auto-generates corrections. You review and approve them.

  1. Go to Brand Intelligence → Alerts
  2. Find the alert (newest at top)
  3. Review the hallucination:
    • What the AI said vs. what’s true
    • Severity and impact
  4. Review the suggested correction (generated from your Truth Nuggets)
  5. Either:
    • Approve: Correction is submitted to the AI engine
    • Modify: Edit the correction before submitting
    • Dismiss: Ignore this hallucination (useful if the AI’s answer is actually more nuanced than your Truth Nugget)
    • Snooze: Re-check in 7 days

Approved corrections are submitted to the AI engine’s feedback mechanisms within minutes and usually take 3-7 days to take effect.

Step 6: Measure Impact with GEO Optimization

If your Brand Trust Score is low, GEO (Generative Engine Optimization) helps improve it.

  1. Go to Brand Intelligence → GEO Audit

  2. Run an audit on your website

  3. Review recommendations sorted by impact:

    • “Add schema.org structured data to pricing page” (+5 points)
    • “Create FAQ for top 5 features” (+4 points)
    • “Update ‘About Us’ with current leadership” (+3 points)

  4. Implement high-impact, low-effort improvements

  5. Re-run the GEO audit in 2-4 weeks to see impact on Brand Trust Score

GEO improvements typically take 2-6 weeks to reflect in AI engine knowledge bases.

Step 7: Track Competitors

Benchmark your brand representation against competitors.

  1. Go to Brand Intelligence → Competitive Benchmarking
  2. Add up to 5 competitors:
    • Competitor name, website, key products
  3. TruthVouch runs the same cross-checks against competitors
  4. Compare:
    • Brand Trust Score (you vs. competitors)
    • Which facts AI engines know about each brand
    • Which competitor is better represented in each engine

Example comparison:

  • ChatGPT knows your company: 85% accurate (pricing wrong by 1 tier)
  • ChatGPT knows Competitor A: 72% accurate (product features outdated)
  • ChatGPT knows Competitor B: 91% accurate (everything correct)

If a competitor is significantly ahead, see what they’re doing:

  • Check their website structure (schema.org usage)
  • Look for recent press coverage
  • Review their FAQ and answer block strategy

Real-World Example

Scenario: You launched a new product last week. ChatGPT doesn’t mention it yet.

  1. Create Truth Nuggets for the new product
    • Name, description, key features, launch date, pricing
    • Link to launch announcement
  2. Activate cross-checks:
    • ChatGPT: Every 1 hour (you want fast adoption)
    • Others: Every 24 hours
  3. Monitor your dashboard
    • Day 1: 0 mentions (ChatGPT hasn’t seen it yet)
    • Day 2: ChatGPT mentions it in 20% of responses
    • Day 3: ChatGPT mentions it in 50% of responses
    • Day 5: ChatGPT mentions it in 80% of responses with 75% accuracy
    • Day 10: ChatGPT mentions it in 95% of responses with 92% accuracy

If adoption is slow (still 0% mentions on day 5):

  • Run GEO audit on your launch announcement page
  • Add schema.org markup for product details
  • Create an FAQ section explaining the product
  • Update your homepage with prominent announcement
  • 2 weeks later, re-check adoption

Metrics to Track

Over time, monitor these metrics:

  • Brand Trust Score: Target 90+. Improving month-over-month?
  • Hallucination count: Trend should be down as corrections take effect
  • Correction resolution time: How long after submitting a correction does the AI fix it? (Usually 3-7 days)
  • AI engine coverage: How many engines mention you? (Target: 9/9)
  • Fact accuracy by category:
    • Company facts: Target 95%+
    • Product facts: Target 85%+
    • Brand claims: Target 80%+
  • Competitive positioning: Brand Trust Score vs. top 3 competitors

Troubleshooting

Brand Trust Score is low (below 60)

  1. Check if Truth Nuggets are accurate and complete (they’re the source of truth)
  2. Run GEO audit to see content gaps AI engines are struggling with
  3. Increase cross-check frequency for a week to get more data
  4. Review top hallucinations to see what patterns exist
  5. Fix patterns (e.g., if pricing is always wrong, ensure your pricing page has clear markup)

Corrections aren’t being applied (AI still says the wrong thing after 2 weeks)

  1. Check if correction was actually submitted (look in alert history)
  2. Try modifying the correction to be more specific or provide a source
  3. Check if the fact in your Truth Nugget is actually correct
  4. Resubmit manually to the AI engine’s feedback system

Brand Trust Score stopped improving

  1. You’ve hit the limit of corrections — now focus on GEO optimization
  2. Run GEO audit and implement recommendations
  3. Add more detailed Truth Nuggets with sources/citations
  4. Check if competitors are doing something better

AI engine not being monitored (no results coming back)

  1. Verify the engine is still operational (try it manually)
  2. Check if your cross-check question is too obscure (try rewording)
  3. Increase cross-check frequency temporarily to get more data points
  4. Contact support if the engine has connectivity issues

Next Steps

Questions? Reach out to your Brand Success Manager or post in the in-app support chat.