Created: June 30, 2025
Your alignment score measures how well AI language models represent your brand attributes across different responses. This comprehensive score averages all subscores from multiple AI models to show your overall brand consistency.
Watch the full demonstration:
How Alignment Scores Work
The alignment score represents the total average of all subscores across different attributes and models. The system tests five alignment prompts from your project settings against multiple AI models, analyzing how well each model's responses align with your defined brand attributes.
When you examine individual results, you'll see responses from various models including GPT, Claude, and others. Each model processes your alignment prompts and generates text responses that get analyzed for brand attribute consistency.
Understanding Your Results
The system evaluates AI responses against specific brand attributes you've defined. For example, if you're testing attributes like "AI for information," "no-code and full-code flexibility," and "user ownership of generated code," the analysis shows percentage similarity between model responses and your desired messaging.
Average alignment scores indicate potential gaps between your brand attributes and available content that AI models access. This suggests either your attributes need refinement or your content strategy requires adjustment to better support your brand messaging.
Source Analysis and Risk Management
The source watchtower feature reveals potential misalignment risks by showing which sources AI models consulted when generating responses. You can identify sources that are misaligned or partially aligned with your brand attributes.
AI models typically access multiple source types including your API documentation, technical docs, website content, and third-party websites. Understanding which external sources influence AI responses helps identify areas for improvement.
Improving Brand Alignment
When you discover misaligned sources, focus on improving content quality, freshness, and messaging consistency. Work with third-party websites that influence AI responses about your brand to ensure accurate representation.
Regular monitoring of alignment scores helps track improvements over time. As you optimize your content and refine your brand attributes, your alignment scores should increase, resulting in more consistent brand representation across AI platforms.
FAQ
What does my alignment score represent?
Your alignment score is the total average of all subscores across different AI models and brand attributes, showing how consistently AI systems represent your brand.
How many prompts does the system test?
The system tests five alignment prompts that you configure in your project settings against multiple AI models.
What sources do AI models access when generating responses?
AI models access your API docs, technical documentation, website content, and various third-party websites when generating brand-related responses.
How can I improve my alignment score?
Improve alignment by refining your content quality, updating messaging consistency, and working with third-party sources to ensure accurate brand representation.
What should I do about misaligned sources?
Review misaligned sources in your source watchtower, check their content accuracy, and work to improve their messaging or reduce their influence on AI responses.