# AI Visibility

> AI Visibility sends questions (direct brand queries + questions generated from the site content) to AI models and measures what they know about your website. You pick keywords and questions in the wizard, 10 questions by default. From every answer we extract brand knowledge, company profile, competitors, sentiment, ranking position and hallucinations.

**Runs with content analysis.** AI Visibility automatically adds the AI SEO Content tool, which determines the main topic, categories and keywords of the site. Discovery questions are generated from them and they also power hallucination detection. The Tech Stack audit is not required; adding it makes the questions more precise about the type of site.

## Three steps

1. **Keywords**: Pick topics, with Google search volume.
2. **Questions**: We draft the questions, you approve or edit them.
3. **Models**: Choose the models and start the run.

## What you get

- **Visibility score**: One number from 0 to 100.
- **Mention rate**: How often AI mentions you.
- **Position**: Where AI ranks you in the answer.
- **Competitors**: Who AI recommends instead of you.
- **Hallucinations**: Where AI makes things up about you.
- **Brand profile**: What AI knows about you and how it rates you.

## Knowledge queries - direct questions about the domain

*2 direct queries × 4 AI models = 8 evaluations*

### K1 - Identity, profile and competition

Does AI know your domain? Complete profile - who is behind it, industry, services/products, target audience, strengths and weaknesses + automatic competitor extraction.

> *“Do you know the website {domain}? Create a profile - who is behind it, what industry it operates in, main products or services, target audience, strengths and weaknesses, and main competitors.”*

### K5 - Recommendation

Would AI recommend your website? Measures the willingness of models to actively recommend.

> *“Would you recommend the website {domain} to someone looking for their type of services or products? Why or why not?”*

## Discovery queries - yours or drafted from the site content, without the domain name

*up to 48 AI-generated queries × 4 models (count depends on site scope)*

### D1 - Main topic

AI generates a question about your main topic without mentioning the domain - tests spontaneous recommendation.

> *“I'm looking for [main topic]. What websites, companies or tools would you recommend?”*

### D2 - Category

Question about your category/segment - does your website appear among the recommendations?

> *“What are the best [category] solutions? I'm mainly interested in [keyword].”*

### D3 - Keyword combination

Question combining your keywords - measures organic visibility.

> *“I need [keyword 1] and [keyword 2]. What platforms or tools do you know?”*

### D4 - Alternative category

Question about a secondary category from enrichment - tests breadth of visibility.

> *“I'm looking for the best [category 2] options. What do you recommend for [keyword]?”*

### D5 - Local market

Question targeting the local market - tests regional visibility.

> *“Where can I find [keyword]? What are the best options on the local market?”*

---

## What is measured in each response

### Knowledge levels

| Level | Score |
|---|---|
| none | 0 pts - AI does not know the domain at all |
| confused | 10 pts - knows it but gets it wrong (hallucination) |
| partial | 40 pts - partial knowledge |
| good | 75 pts - good knowledge, correct profile |
| excellent | 100 pts - excellent, complete profile |

### Collected data

- **Company profile** - type, industry, services, target audience, region
- **Reputation** - sentiment (positive/neutral/negative), strengths and weaknesses
- **Competitors** - automatically extracted domains with mention frequency
- **Ranking position** - at which position AI recommends your domain (1-5+)
- **Direct link** - whether the answer contains a direct HTTP link to the website
- **Model confidence** - how confident the model is (0.0-1.0)

## Hallucination detection (cross-check)

We compare each AI model response against real data from the content analysis: the main website topic and keywords. If the AI response doesn't match what the website actually does, we detect a hallucination and reduce the score.

- **Similarity ≥ 0.50** - response consistent, no score change
- **Similarity 0.35-0.49** - partial match, knowledge level capped at "partial"
- **Similarity < 0.35** - hallucination! Downgraded to "confused"

Example: Model claims "IT circular economy" but fingerprint shows "moving company" → similarity 0.31 → hallucination detected, score reduced.

## Visibility scoring

Base score = knowledge level (none=0, confused=10, partial=40, good=75, excellent=100)

**Bonuses:**

- +5 - response contains a direct link to the domain
- +5 - model confidence ≥ 0.8
- +5 - positive reputation sentiment

**Penalties:**

- -10 - response processing failed (fallback mode)

**Overall visibility score = 25% knowledge score + 75% discovery score. Discovery = 70% mention rate + 30% position score.**

## Tested AI models (4)

- ChatGPT (OpenAI)
- Google Gemini (Google)
- Perplexity Sonar (Perplexity AI)
- Grok (xAI)

Each query is tested on all 4 models simultaneously. Results are aggregated by query, model and topic.