# How to Win AI Citations for Your Brand

Answer Spy in QueryBurst probes AI models with strategic questions from multiple angles — with live search grounding — and extracts the criteria that appear repeatedly across responses: trust signals, expertise indicators, pricing transparency, customer evidence, certifications. These criteria are consistent within a category but not publicly documented. Use it to see exactly what AI values when recommending businesses in your industry, ranked by confidence with category breakdowns.

## How QueryBurst Answer Spy Works

### The Pipeline

1. **Generate Probes** - Creates 5 strategic prompts approaching your topic from different angles  
2. **Query AI** - Sends probes to an LLM with Search grounding for real-world context  
3. **Extract Criteria** - Analyzes each response to identify evaluation factors  
4. **Synthesize** - Merges criteria across responses, calculates confidence, categorizes

### Probe Angles

Each analysis generates probes covering:

| Angle                | What It Probes                       |
|---------------------|--------------------------------------|
| **Selection criteria** | What to look for when choosing      |
| **Comparisons**       | Best options, top providers, rankings |
| **Red flags**         | What to avoid, warning signs        |
| **Questions to ask**  | Consultation preparation             |
| **Costs/fees**       | Pricing, what's fair                |

## Analysis Types

### Service Analysis

For service providers (attorneys, plumbers, consultants, etc.)

**Fields:**

- Service type (required)
- Location - Geographic targeting
- Specialization - Niche focus
- Target customer - Who you serve

### Product Analysis

For products (SaaS, ecommerce, physical goods)

**Fields:**

- Product description (required)
- Region/Market - Geographic scope
- Key use case - Primary function
- Target user - Ideal customer profile

### Content Analysis

For content sites (guides, reviews, how-tos)

**Fields:**

- Content topic (required)
- Region/Audience - Geographic or demographic
- Content type - Format (articles, videos, etc.)
- Target reader - Who you're writing for

## Understanding Results

### Criteria Tab

Each extracted criterion includes:

| Element        | Description                                   |
|----------------|-----------------------------------------------|
| **Criterion**  | The specific factor AI evaluates              |
| **Category**   | Grouping (Expertise, Trust, etc.)            |
| **Confidence**  | How consistently this appeared                |
| **Frequency**  | How many responses mentioned it                |
| **Recommendation** | How to address this criterion               |

### Confidence Levels

- 🟢 **High** - Appeared in most responses, core factor  
- 🟡 **Medium** - Appeared in some responses, important  
- ⚫ **Low** - Appeared once, may be edge case

### Categories

Criteria are grouped into categories like:
- **Expertise & Experience** - Qualifications, track record  
- **Trust & Credibility** - Reviews, certifications, transparency  
- **Service Delivery** - Communication, process, availability  
- **Value** - Pricing, outcomes, ROI

### AI Responses Tab

View the raw AI responses that criteria were extracted from:
- Full response text with markdown formatting  
- Response length and word count

## Investigate Site

After extracting criteria, you can investigate how well your site covers them.

### How Investigation Works

1. **Search** - AI searches your indexed pages for relevant content  
2. **Read** - Reads content sections looking for each criterion  
3. **Report** - Reports coverage with evidence and recommendations

### Coverage Levels

- ✅ **Strong** - Clear, prominent coverage found  
- ⚠️ **Weak** - Partial or buried coverage  
- ❌ **Missing** - No relevant content found

### Investigation Results

Each criterion shows:
- Coverage status (strong/weak/missing)  
- Evidence quotes from your content  
- Page URLs where evidence was found  
- Recommendations for improvement

## Best Practices

### Getting Better Results

✅ **Do** be specific about your service/product  
✅ **Do** include location for local services  
✅ **Do** fill in advanced options for more targeted criteria  
✅ **Do** run multiple analyses for different niches

❌ **Don't** use overly broad terms ('consultant')  
❌ **Don't** skip location for location-sensitive services  
❌ **Don't** expect identical results each run (AI varies)

### Acting on Criteria

**High Confidence Criteria:**
- These are your priority - AI consistently looks for these  
- Create dedicated content addressing each  
- Make coverage prominent and easy to find

**Medium Confidence Criteria:**
- Important differentiators  
- Address these to stand out from competitors  
- May be context-dependent (location, specialty)

**Low Confidence Criteria:**
- Consider but don't prioritize  
- May become more important over time  
- Good for comprehensive coverage

### Improving Coverage

After investigation:
1. **Missing criteria** → Create new content  
2. **Weak coverage** → Strengthen existing content  
3. **Strong coverage** → Maintain and update

## Timing

- **Criteria Analysis**: 1-2 minutes  
- **Site Investigation**: 1-2 minutes  
- **Parallel queries**: 5 AI requests run simultaneously

## Technical Notes

### Grounding

Probes are sent to an LLM with Search grounding, meaning responses include real-world search data, not just model knowledge.

### Synthesis

Criteria appearing in multiple responses are merged, with confidence calculated based on frequency and consistency. Similar criteria are deduplicated.