Trying to keep up with AI search terminology can be overwhelming. With new platforms, tools, and best practices emerging rapidly, even experienced marketers and technical teams find themselves searching for clear answers. Scrunch is built for the next era of search—so it’s only fitting that we help demystify the language powering this transformation.
That’s why we created the Scrunch AI Glossary. Here you’ll find simple definitions for the most important terms in AI-driven search, content optimization, and Scrunch’s own platform features. Bookmark this page whenever you need to quickly clarify an AI or Scrunch-specific buzzword.
A comprehensive list of Scrunch and AI search definitions
Here are all the must-know definitions to make understanding AI optimization and Scrunch terminology a bit more manageable:
AI Optimization
The practice of ensuring website content is readable, understandable, and favored by AI platforms like ChatGPT, Perplexity, and Google AI Overviews. This includes technical accessibility, contextual clarity, and writing style adjustments.
Presence
A measurement of how often your brand or domain is cited or mentioned in AI-generated answers across platforms.
Competitive Presence
This metric compares your brand's representation across AI responses to that of your competitors. A high score means your brand is being mentioned or cited more frequently than peers in responses to similar prompts.
Position Score
An evaluation of where your brand appears in an AI-generated response:
Top = Your brand is mentioned or cited in the top 25% of the response
Middle = Your brand is mentioned or cited in the middle 50% of the response
Bottom = Your brand is mentioned or cited in the bottom 25% of the response
Higher placements are typically more visible to users and indicate stronger AI relevance.
We're working to expand support for position metrics. Right now, you can view position metrics within each prompt response, but you cannot view or filter by position metrics on the dashboard or prompts tab.
Sentiment Score
A qualitative measure of how your brand is being described in AI responses:
Positive = Language is favorable or endorsing
Mixed = Includes pros and cons or neutral observations
Negative = Highlights drawbacks or criticism
This helps gauge brand perception in AI-generated answers.
Context Tab
A Scrunch feature where you can define your brand’s key information: personas, competitors, and topics. This data powers prompt generation and segmentation.
Customer Persona
A fictionalized profile of a target customer. Used in Scrunch to tailor prompt generation based on different user intents.
Key Topics
Themes or concepts you want to monitor across prompts. Used for filtering and analyzing presence across content areas.
Prompts
The natural language questions or statements you're tracking in Scrunch to see how AI models respond. Each prompt is counted once per model (e.g., once for ChatGPT, once for Perplexity). Prompts form the backbone of your monitoring configuration.
Prompts (metric on the Prompts tab)
A metric showing how prompts are being tracking in Scrunch (each AI platform the prompts is running against counts as a single prompt toward the total).
Responses
The total number of AI-generated responses captured by Scrunch for your prompts. Scrunch regularly re-runs your tracked prompts across supported models and stores the full outputs. This count shows how much real-world data you've already accumulated for analysis.
Training Data Crawlers
Bots like GPTBot, ClaudeBot, or CCBot that collect data for AI model training. You can block these in robots.txt without hurting AI citation performance.
Bot Blocking
A configuration that restricts automated bots (like those used by AI or search engines) from accessing your website content. Over-aggressive bot blocking can prevent AI platforms from citing your pages.
AI User Agent
A bot identifier used by AI platforms when retrieving web content. Examples include ChatGPT-User, PerplexityBot, and meta-externalagent. Allowlisting these in robots.txt and bot protection tools ensures visibility in AI responses.
PerplexityBot
Perplexity AI’s bot that handles both search indexing and real-time content retrieval. Should be allowlisted for presence in Perplexity responses.
ChatGPT-User
OpenAI’s bot used to retrieve web content in real-time for ChatGPT answers. Must be allowlisted for your site to be cited in ChatGPT responses.
AI Traffic tab
Website visits referred from AI platforms. A Looker Studio embedding - This is visible for orgs who have existing GA4 environments and are currently logged in to a Google account with access to those environments.
Site Audit tab
A Scrunch tool that analyzes how your site appears to AI bots. Checks for accessibility, context quality, performance, and formatting.
Access Controls Score
A category within the Scrunch Site Audit that evaluates whether AI bots (like ChatGPT-User or PerplexityBot) can access your site. This includes checks on robots.txt, firewall settings, and bot protections (e.g., Cloudflare). A low score often indicates technical barriers that could block AI from citing your content.
Content Clarity
A metric in Scrunch’s Site Audit that assesses how clearly AI models can interpret your page after stripping out styling and JavaScript. Pages with overly complex structures, fragmented markup, or text buried in visual elements tend to perform poorly.
Does your robots.txt have specific rules allowing all key AI bots?
A proactive Site Audit flag that checks whether your robots.txt file explicitly allows AI user agents like ChatGPT-User, PerplexityBot, and meta-externalagent. While not required for access, allowing these bots is recommended to reduce friction and avoid blocking AI visibility unintentionally.
Dynamic Content
Website content loaded via JavaScript. Most AI crawlers cannot process this, which means dynamically rendered content may not be cited unless pre-rendered or supported with fallback text.
Structured Data (Schema)
Metadata like JSON-LD that can enrich SEO results. For AI, however, the actual on-page text content is far more influential. Schema should complement - not replace - clear text.