Skip to main content

​
What is analysis?

Analysis is the process of evaluating AI provider responses to extract actionable brand intelligence. OneGlanse analyzes each response to measure:
  • Presence - Is your brand mentioned? How prominently?
  • Position - Where do you rank in recommendation lists?
  • Sentiment - How favorably is your brand portrayed?
  • Recommendation - Does the AI actively suggest your brand?
  • Competition - Who else appears and how do you compare?
  • Perception - What narrative is the AI building about your brand?
  • Risks - Are there factual errors or negative associations?
This multi-dimensional analysis transforms raw AI responses into strategic insights.

​
Why analysis matters

AI responses shape user perceptions and purchasing decisions. Analysis helps you: Measure AI visibility - Quantify how often and how prominently AI mentions your brand Track competitive positioning - Understand where you rank against alternatives Detect sentiment shifts - Catch changes in how AI describes your brand before they impact business Identify content gaps - Find areas where AI lacks current information about your brand Guide content strategy - Prioritize creating content that improves AI understanding Validate messaging - Check if AI responses reflect your brand positioning

​
How analysis works

​
Analysis structure

Each prompt response generates a comprehensive BrandAnalysisResult: 
interface BrandAnalysisResult {
  // High-level metadata
  metadata?: {
    brandName: string;
    brandDomain: string;
    prompt: string | null;
    prompt_id: string | null;
    analyzedAt: string;
  };
  
  // Composite score (0-100)
  geoScore: {
    overall: number;
    verdict: string;
  };
  
  // Six core analysis dimensions
  presence: PresenceMetrics;
  position: PositionMetrics;
  sentiment: SentimentMetrics;
  recommendation: RecommendationMetrics;
  competitors: CompetitorMetrics[];
  perception: PerceptionMetrics;
  
  // Actionable insights
  risks: RiskMetrics;
  actions: ActionItem[];
}

​
Analysis dimensions

The GEO Score (Generative Engine Optimization Score) is a composite 0-100 score that answers: “How well is your brand performing in AI responses?”
geoScore: {
  overall: 75,  // 0-100 composite score
  verdict: "Strong visibility with room for improvement in competitive positioning"
}
Higher scores indicate better overall performance across all dimensions.
Measures whether and how prominently your brand appears:
presence: {
  mentioned: true,           // Brand appears in response
  mentionCount: 3,           // Number of times mentioned
  visibility: 85,            // 0-100 prominence score
  prominence: "significant", // Classification level
  firstMentionPosition: "top" // Where brand first appears
}
Prominence levels:
  • dominant - Primary focus of the response
  • significant - Major part of the discussion
  • moderate - Notable but not central
  • minor - Brief mention
  • passing - Barely referenced
  • absent - Not mentioned
Position values:
  • top - First third of response
  • middle - Second third
  • bottom - Final third
  • absent - Not mentioned
Tracks your ranking in recommendation or comparison lists:
position: {
  rankPosition: 2,           // Position in list (null if unranked)
  totalRanked: 5,            // Total items in list
  isTopPick: false,          // Marked as #1 recommendation
  isTopThree: true,          // In top 3 recommendations
  rankingContext: "Best project management tools for remote teams"
}
This reveals competitive standing when AI lists multiple options.
Analyzes the tone and favorability of brand mentions:
sentiment: {
  score: 72,                 // 0-100 (negative to positive)
  label: "positive",         // Classification
  positives: [
    "Excellent collaboration features",
    "Intuitive user interface",
    "Strong mobile app"
  ],
  negatives: [
    "Higher pricing than competitors",
    "Steeper learning curve for advanced features"
  ]
}
Sentiment labels:
  • very_positive (80-100) - Highly favorable
  • positive (60-79) - Generally favorable
  • neutral (40-59) - Balanced or factual
  • negative (20-39) - Unfavorable elements
  • very_negative (0-19) - Significantly negative
Evaluates how strongly the AI recommends your brand:
recommendation: {
  type: "strong_alternative",  // Recommendation strength
  bestFor: [
    "Teams needing visual project tracking",
    "Organizations using agile methodologies"
  ],
  caveats: [
    "May be expensive for small teams",
    "Requires time investment to learn advanced features"
  ]
}
Recommendation types:
  • top_pick - Primary recommendation
  • strong_alternative - Highly recommended option
  • conditional - Recommended for specific use cases
  • mentioned_only - Listed but not actively recommended
  • discouraged - Mentioned with warnings
  • not_mentioned - Absent from response
Identifies competing brands in the response:
competitors: [
  {
    name: "Asana",
    domain: "asana.com",
    visibility: 90,          // Prominence score
    sentiment: 85,           // Favorability score
    rankPosition: 1,         // List position
    isRecommended: true,     // Actively recommended
    winsOver: ["Trello", "Basecamp"],
    losesTo: []
  },
  {
    name: "Monday.com",
    domain: "monday.com",
    visibility: 75,
    sentiment: 78,
    rankPosition: 3,
    isRecommended: true,
    winsOver: ["Basecamp"],
    losesTo: ["Asana"]
  }
]
This reveals the competitive landscape in each AI response.
Extracts how the AI characterizes your brand:
perception: {
  coreClaims: [
    "Comprehensive project management platform",
    "Strong integration ecosystem",
    "Enterprise-grade security"
  ],
  differentiators: [
    "Advanced automation capabilities",
    "Customizable workflows"
  ],
  bestKnownFor: "Visual project tracking with timeline views",
  pricingPerception: "premium"
}
Pricing perceptions:
  • premium - Positioned as high-end
  • mid_range - Standard pricing
  • budget - Affordable option
  • free - Free tier or freemium model
  • not_mentioned - No pricing discussion

​
Risk detection

Analysis automatically identifies potential issues: 
risks: {
  hasRisks: true,
  items: [
    {
      type: "outdated_info",
      severity: "warning",
      detail: "Response mentions version 2.0 but current version is 3.5"
    },
    {
      type: "factual_error",
      severity: "critical",
      detail: "Claims brand doesn't support mobile, but mobile apps exist"
    },
    {
      type: "negative_association",
      severity: "warning",
      detail: "Brand associated with 'complex setup' multiple times"
    }
  ]
}
Risk types:
  • outdated_info - AI has old information
  • factual_error - Incorrect claims about your brand
  • brand_confusion - Mixed up with another brand
  • negative_association - Linked with unfavorable terms
  • missing_from_response - Expected to appear but didn’t
Severity levels:
  • critical - Immediate attention required
  • warning - Should be addressed
  • info - Worth monitoring

​
Actionable recommendations

Each analysis includes prioritized action items: 
actions: [
  {
    priority: "high",
    recommendation: "Create content highlighting the new mobile app to update AI knowledge"
  },
  {
    priority: "medium",
    recommendation: "Develop comparison guides against Asana to improve competitive positioning"
  },
  {
    priority: "low",
    recommendation: "Monitor sentiment trend for 'setup complexity' over next 30 days"
  }
]

​
Analysis storage

Analysis results are stored in ClickHouse for efficient querying: 
interface PromptAnalysis {
  id: string;
  prompt_id: string;
  workspace_id: string;
  user_id: string;
  model_provider: string;
  prompt: string;
  brand_analysis: string;  // JSON-stringified BrandAnalysisResult
  prompt_run_at: string;
  created_at: string;
}
This structure enables:
  • Time-series queries - Track metrics over time
  • Provider comparisons - Compare performance across AI platforms
  • Prompt analysis - See which prompts generate best visibility
  • Competitive tracking - Monitor competitor mentions over time

​
Analysis records

The flattened AnalysisRecord structure combines response and analysis data: 
interface AnalysisRecord {
  // Identifiers
  id: string;
  prompt_id: string;
  prompt_run_at: string;
  prompt: string;
  
  // Context
  user_id: string;
  workspace_id: string;
  model_provider: string;
  
  // Data
  response: string;
  sources: Source[];
  brand_analysis?: BrandAnalysisResult;
  
  // Status
  is_analysed?: boolean;
  
  // Timestamp
  created_at: string;
}
This enables efficient filtering and aggregation across multiple dimensions.

​
Interpreting analysis results

​
Single response analysis

For a single prompt/provider combination, analysis reveals:
1

Brand presence

Did the AI mention your brand? If yes, how prominently?
// Example: Strong presence
{
  mentioned: true,
  mentionCount: 4,
  visibility: 85,
  prominence: "significant"
}
2

Competitive context

Who else was mentioned and how do you compare?
// Example: You're #2 of 5
{
  rankPosition: 2,
  totalRanked: 5,
  isTopThree: true,
  competitors: ["Competitor A", "Competitor B", ...]
}
3

Sentiment quality

What’s the tone of the mention?
// Example: Positive with caveats
{
  score: 72,
  label: "positive",
  positives: ["Great features", "Easy to use"],
  negatives: ["Higher price point"]
}
4

Risk identification

Are there any issues to address?
// Example: Outdated information
{
  hasRisks: true,
  items: [{
    type: "outdated_info",
    severity: "warning",
    detail: "Mentions discontinued feature"
  }]
}

​
Aggregate analysis

Across multiple prompts and providers, track: Mention rate - What percentage of responses include your brand? 
// Example: 65% mention rate
const mentionRate = mentionedCount / totalResponses;
// 13 mentions across 20 responses = 65%
Average visibility - How prominently do you appear on average? 
// Example: 68/100 average visibility
const avgVisibility = sum(visibility scores) / mentionedCount;
Sentiment trend - Is sentiment improving or declining? 
// Example: Positive trend
[
  { date: "2026-03-01", sentiment: 65 },
  { date: "2026-03-08", sentiment: 70 },
  { date: "2026-03-15", sentiment: 73 }
]
Position distribution - How often do you rank #1, #2, #3, etc.? 
// Example: Position breakdown
{
  "1": 3,  // Top pick 3 times
  "2": 5,  // Second place 5 times
  "3": 4,  // Third place 4 times
  "null": 8 // Not ranked 8 times
}

​
Analysis metadata

Metadata provides context for filtering and exploration: 
interface AnalysisMetadata {
  available_brands: [
    { name: "Your Brand", website: "yourbrand.com" },
    { name: "Competitor A", website: "competitor-a.com" }
  ],
  available_models: [
    "chatgpt",
    "perplexity",
    "gemini",
    "ai-overview"
  ]
}
This enables dynamic filtering in the UI:
  • Filter by specific competitor
  • Compare performance across providers
  • Track brand mentions over custom time periods

​
Analysis API response

The complete analysis API returns: 
interface AnalysisResponse {
  records: AnalysisRecord[];     // Individual response analyses
  metadata: AnalysisMetadata;     // Available filters
}
This powers the dashboard views, trend charts, and competitive analysis features.

​
Using analysis for strategy

​
Content prioritization

Use analysis to guide content creation:
  1. High-risk items - Address factual errors and outdated information immediately
  2. Low-visibility topics - Create content for prompts where you’re rarely mentioned
  3. Competitive gaps - Develop comparison content for areas where competitors dominate
  4. Sentiment improvements - Address recurring negative associations

​
Tracking campaigns

Monitor analysis metrics before and after content campaigns: 
// Before campaign
{
  mentionRate: 0.45,      // 45% of responses
  avgVisibility: 58,      // Moderate prominence
  avgSentiment: 65,       // Positive
  avgPosition: 3.2        // Usually #3
}

// After campaign (2 months later)
{
  mentionRate: 0.72,      // 72% of responses (↑ 27%)
  avgVisibility: 78,      // High prominence (↑ 20)
  avgSentiment: 74,       // More positive (↑ 9)
  avgPosition: 2.1        // Usually #2 (↑ 1.1)
}

​
Competitive intelligence

Analyze competitor metrics to understand market positioning:
  • Which competitors appear most frequently?
  • What are their key differentiators according to AI?
  • Where do they rank relative to your brand?
  • What’s their sentiment score?
This intelligence informs competitive strategy and messaging.

Interpreting Metrics

Detailed guide to understanding and acting on analysis data

Prompts

Learn how to create prompts that generate useful analysis

Providers

Understand provider-specific analysis differences

API Reference

Technical documentation for the analysis API

Build docs developers (and LLMs) love

Get started for freeTalk to us