Guides

Brand Tracking Guide

Learn how to effectively track your brand and competitors across AI platforms

Overview

Brand tracking in JHorizon helps you understand how AI models perceive and recommend your brand compared to competitors. This guide covers best practices for setting up effective tracking.

Setting Up Your Brands

Add Your Primary Brand

During organization setup, configure your main brand:

  • Brand Name: Your official company/product name
  • Category: Industry or product category
  • Description: What your brand does (used for context)

The brand description helps AI models understand context when analyzing mentions.

Add Competitors

Track up to 5 competing brands:

  1. Choose direct competitors in your space
  2. Use their official brand names
  3. Focus on brands AI models are likely to know

Select competitors mentioned frequently in your industry to get meaningful comparison data.

Configure Tracked Queries

Choose prompts that:

  • Represent real user questions
  • Target your specific use cases
  • Cover different angles (features, pricing, alternatives)

Examples:

  • "What are the best AI analytics tools?"
  • "Compare brand tracking platforms"
  • "Alternatives to [Competitor]"

Types of Tracked Queries

Recommendation Queries

Goal: Measure if AI models recommend your brand

Examples:

  • "What's the best [category] tool?"
  • "Top 5 [category] platforms"
  • "Which [category] should I choose?"

Comparison Queries

Goal: Understand competitive positioning

Examples:

  • "Compare [YourBrand] vs [Competitor]"
  • "Differences between [brands]"
  • "[YourBrand] or [Competitor] for [use-case]?"

Alternative Queries

Goal: Track substitution patterns

Examples:

  • "Alternatives to [Competitor]"
  • "[Competitor] competitors"
  • "Similar tools to [AnyBrand]"

Feature Queries

Goal: Assess feature recognition

Examples:

  • "Which [category] has [feature]?"
  • "Best [category] for [use-case]"
  • "[Feature] in [category] tools"

Query Best Practices

Do's

Use natural language - Write queries as users would ask them ✅ Cover multiple angles - Recommendations, comparisons, alternatives ✅ Be specific - Target your actual use cases and customer questions ✅ Include context - Specify industry, user type, or requirements ✅ Test variations - Try different phrasings of similar questions

Don'ts

Avoid brand stuffing - Don't force your brand name into every query ❌ Skip generic queries - Too broad = less actionable insights ❌ Ignore user intent - Track what users actually search for ❌ Overload categories - Focus on your core positioning ❌ Use outdated terms - Keep queries aligned with current market language

Understanding Rankings

JHorizon extracts brand rankings from LLM responses:

Ranking Metrics

Position: Where your brand appears (1 = first mentioned)

Mention Rate: Percentage of responses that mention your brand

Average Rank: Mean position across all mentions

Share of Voice: Relative mention frequency vs competitors

Rankings are extracted per response, then aggregated for weekly analytics.

Sentiment Analysis

Each brand mention is analyzed for sentiment:

  • Positive: Favorable description, recommendation, or praise
  • Neutral: Factual mention without positive/negative context
  • Negative: Criticism, limitation, or unfavorable comparison

Sentiment Score vs. Sentiment Contribution

JHorizon surfaces two distinct sentiment metrics. They answer different questions and are calculated differently — pay attention to which one you're looking at.

Sentiment Score (top card)

Question it answers: "When your brand IS mentioned, how positive is the tone?"

This is a pure tone metric. It does not matter whether you are mentioned 1 time, 100 times, or 5 times out of a thousand — it only measures the overall sentiment of the mentions that exist.

A brand mentioned once with glowing praise will score the same as a brand mentioned 100 times with the same average tone.

Sentiment Contribution (GEO score dimension)

Question it answers: "How much does sentiment actually contribute, given how often the brand appears?"

This metric weights tone by how often the brand actually shows up. If you are mentioned only 1 out of 100 times, your score is bumped lower — even with perfect sentiment on that single mention, it barely contributes to the overall GEO score because there is so little signal behind it.

This is why Sentiment Contribution appears as a dimension inside the Overall GEO Score: it reflects sentiment's real-world impact on your brand visibility, not just the tone of isolated mentions.

Rule of thumb: Sentiment Score tells you how favorably you are spoken about. Sentiment Contribution tells you how much that favorability is actually moving the needle on your GEO score.

Improving Sentiment

Identify Negative Patterns

Look for recurring themes in negative mentions:

  • Pricing concerns
  • Missing features
  • Competitive weaknesses

Analyze Citation Sources

Check which websites AI models reference:

  • Are they outdated?
  • Do they misrepresent your brand?
  • Can you provide better sources?

Update Public Information

Ensure your:

  • Official website is current
  • Documentation is comprehensive
  • Case studies are published
  • Press releases highlight strengths

Citation Analysis

JHorizon scrapes and analyzes the sources AI models cite:

What to Track

Source Domains: Which websites get referenced most

Content Recency: How old are the cited pages

Brand Context: How your brand appears in citations

Competitor Context: How competitors appear in same sources

Optimizing Citations

AI models often cite authoritative sources like industry publications, review sites, and official documentation.

Strategies:

  1. Get Featured: Publish on high-authority sites
  2. Update Content: Keep your web presence current
  3. Create Resources: Comprehensive guides and documentation
  4. Earn Reviews: Encourage customers to post on review platforms
  5. Publish Data: Original research and statistics

Multi-Model Insights

JHorizon tracks 5 AI models:

  • OpenAI (GPT-4, GPT-4 Turbo)
  • Anthropic (Claude)
  • Google (Gemini)
  • Meta (Llama)
  • Others (as added)

Model Differences

Different models may:

  • Rank brands differently
  • Use different citation sources
  • Show varying sentiment
  • Update knowledge at different rates

Compare model performance to identify knowledge gaps or opportunities.

Weekly Analytics

Workflows run weekly to track changes over time:

Week-over-Week Growth:

  • Mention rate changes
  • Ranking improvements/declines
  • Sentiment shifts

Seasonal Patterns:

  • Industry trends
  • Product launch impacts
  • Competitive activity

Long-Term Progress:

  • Brand awareness growth
  • Market position evolution
  • Citation diversity

Next Steps