Sentinel Query Builder
Build your strategic query set for AI citation tracking using the Five-Pillar Query Architecture.
The Three Streams GEO Methodology
Educational Use Only — No Responsibility Assumed
This tool is provided for educational purposes only as part of the Three Streams GEO Methodology. The creators assume no responsibility for business decisions or outcomes resulting from use of this tool. Query selection should be validated against actual customer search behavior and business priorities.
What Are Sentinel Queries?
Sentinel queries are a defined set of 50-75 strategic queries representing your customer journey across AI platforms. They provide consistent, repeatable measurement of citation performance through ACF (AI Citation Frequency) and SOV-AI (Share of Voice in AI).
The Five-Pillar Architecture
Research suggests 50-75 queries balances comprehensive coverage with manageable tracking. Below 50 provides insufficient coverage; above 100 shows diminishing returns.
- Branded: Direct brand recognition
- Problem: User problem identification
- Solution: Solution-seeking behavior
- Competitive: Comparative positioning
- Product: Specific product visibility
Distribution across pillars varies based on your strategic context. Select a model below.
⚠️ Risk Detection Queries (Separate Tracking Set)
15-20 queries (~20% of your general sentinel count) maintained as a separate tracking set from performance queries. These serve an early warning function where the desired outcome is often absence of problematic mentions.
- Information Void: AI hedging or citing competitors instead of you
- Citation Decay: AI citing outdated information
- Training Contamination: AI surfacing negative sentiment
- Authority Erosion: Competitors cited instead of your brand
- Technical Accessibility: AI can't find info that exists on your site
Key Difference: Success = absence of problematic mentions (unlike performance queries where presence is success). Tracked in separate dashboard with different success criteria.
Your Brand Information
Query Distribution
Select a distribution model based on your brand's strategic context, then adjust total query count.
💡
Equal Distribution
Equal distribution across all five categories provides a balanced baseline when strategic priorities are unclear, when establishing initial benchmarks, or when your brand is mid-maturity without extreme strengths or weaknesses.
Primary Goal: Balanced measurement across all intent categories.
| Category | Allocation | Queries | Rationale |
|---|
75
Branded (15)
Problem (15)
Solution (15)
Competitive (15)
Product (15)
Pillar 1
Branded Queries
0/15
Purpose: Understand how AI systems describe your brand when specifically asked. These queries measure direct brand recognition and sentiment.
Example Patterns (click to add)
Brand reviews
Is Brand good
Brand vs Competitor
Where to buy
Brand warranty
Brand quality
Known for
Pillar 2
Problem Recognition Queries
0/15
Purpose: Capture users identifying problems your products solve. These queries represent the "Awareness" stage of the customer journey.
Example Patterns (click to add)
Why is my...
What causes...
How to prevent...
Signs of...
Is X normal
Why does X happen
Pillar 3
Solution Seeking Queries
0/15
Purpose: Target users actively seeking solutions. These queries represent the "Consideration" stage where users evaluate options.
Example Patterns (click to add)
How to...
Best way to...
Tips for...
What helps with...
How to fix...
What should I use
Pillar 4
Competitive Comparison Queries
0/15
Purpose: Measure your competitive positioning in AI responses. These queries show where you stand against competitors.
Example Patterns (click to add)
Best in category
Brand vs Competitor
Category comparison
Top brands
Category rankings
Which should I buy
Pillar 5
Product-Specific Queries
0/15
Purpose: Measure visibility for specific product types and use cases. These queries target users with defined needs.
Example Patterns (click to add)
Product for type
Professional product
Feature + product
Product under price
Product for use case
Product for beginners
⚠️ Risk Detection Queries
Separate tracking set for early warning — success = absence of problematic mentions
0 queries
Target: 15-20 (~20% of performance queries)
Risk 1
Information Void
0/4
Risk Signal: AI hedges, cites competitors, or provides inaccurate information instead of your brand content.
Example Patterns (click to add)
Source questions
Policy questions
Process questions
Ingredient questions
Risk 2
Citation Decay
0/4
Risk Signal: AI cites outdated information or old publication dates instead of current content.
Example Patterns (click to add)
Current pricing
Discontinued items
Latest products
Current policies
Risk 3
Training Contamination
0/4
Risk Signal: AI surfaces negative sentiment or community complaints that have been embedded in training data.
Example Patterns (click to add)
General sentiment
User opinions
Review queries
Trust queries
Risk 4
Authority Erosion
0/4
Risk Signal: AI cites competitors or third parties rather than your brand sources for category queries.
Example Patterns (click to add)
Category queries
Comparison queries
Category leaders
Direct comparison
Risk 5
Technical Accessibility
0/4
Risk Signal: AI cannot find information that exists on your website (indicates crawler or rendering issues).
Example Patterns (click to add)
Product specs
Ingredient lists
Policy info
Usage instructions
Sentinel Query Set
75 queries across 5 pillars
Branded
0
Problem
0
Solution
0
Competitive
0
Product
0
📊 Tracking Guidance
| Platform | Tracking Frequency | What to Document |
|---|---|---|
| ChatGPT | Weekly | Appearance, position, competitors mentioned, source cited |
| Perplexity | Weekly | Appearance, position, competitors mentioned, source cited |
| Claude | Weekly | Appearance, position, competitors mentioned, source cited |
| Google Gemini / AIO | Weekly | Appearance, position, competitors mentioned, source cited |