The Core Concept: Wins vs. Losses
Every time a tracking run executes, your prompts are sent to multiple AI engines (ChatGPT, Gemini, Perplexity, Google AI). Each response is then evaluated:- Win (Mentioned) - AI mentioned your brand in its answer
- Loss (Not Mentioned) - AI did not mention your brand. It mentioned competitors or other brands instead
What You See
Win Rate
Your headline metric showing how often AI mentions your brand across all analyzed prompts.| Metric | What It Means |
|---|---|
| Mentioned (green) | Number of prompts where AI mentioned your brand |
| Not Mentioned (red) | Number of prompts where AI did not mention your brand |
| Win Rate | Mentioned divided by Total, shown as a percentage |
Why You Win
When your brand IS mentioned, the analysis identifies what is working in your favor. These are your competitive advantages - the things you should continue doing. Examples include:- Strong positioning as the recognized leader in your space
- High perceived relevance and clarity of what you offer
- Brand name that directly matches common search queries
Key Gaps to Address
When your brand is NOT mentioned, the analysis pinpoints what is missing. These are your action items. Examples include:- No dedicated page explicitly addressing a specific query topic
- Missing third-party discussions and endorsements
- Lack of structured content that AI can easily retrieve
How It Works
Every tracking run sends your prompts to multiple AI engines and captures the full response, including which brands were mentioned, how many times, and which websites were cited.
The system prioritizes responses where competitors beat you, followed by emerging market opportunities, and then your wins. Every prompt gets at least one analysis so nothing is skipped entirely.
For wins, the system identifies your strengths and what you should keep doing. For losses, it identifies what the winning competitor did differently and what specific improvements you should make.
The same prompt may be analyzed across multiple tracking runs. Only the latest analysis per prompt is kept, so you always see the most current competitive picture without inflated numbers.
How Wins and Losses Are Determined
A “win” or “loss” is determined by who got the most mentions or citations in each AI response:- Mentions Win - Your brand name appeared more times than any competitor in the AI response
- Citations Win - The AI cited your website more than any competitor’s website in its sources
Coverage increases with each tracking run. If you recently added new prompts, they will be included in the next analysis cycle. Over time, every prompt in your workspace gets analyzed.
How to Improve Your Win Rate
Quick Wins (days to weeks)
Quick Wins (days to weeks)
- Update your Google Business Profile - Complete all fields, add photos, and respond to reviews
- Add FAQ pages that directly answer your tracked prompts
- Include statistics and data points on your website such as room counts, ratings, and distances to landmarks
Medium-Term (weeks to months)
Medium-Term (weeks to months)
- Get cited by authoritative sources - Tourism boards, review sites, and industry publications
- Create content with quotes and testimonials that AI can reference
- Use proper structured data markup so AI can easily extract information about your brand
Long-Term (months)
Long-Term (months)
- Build brand mentions across the web - Forums, social media, and press coverage
- Publish original research or data that AI models will pick up over time
- Maintain consistent content updates - Freshness signals matter to AI engines
Understanding Confidence Levels
Each analysis comes with a confidence score from 0 to 100%. Higher confidence means the system is more certain about why you won or lost.- 80%+ - Very reliable findings
- 60-80% - Solid directional guidance
- Below 60% - Treat as directional hints rather than certainties
How Often Is It Updated?
- Gap analysis runs automatically after each tracking run
- The briefing always shows the latest result per prompt
- Running more tracking runs means more prompts get analyzed and better overall coverage
- Results are cached for 7 days to avoid redundant analysis of the same patterns