AI Signal

What you receive from a diagnostic

A single answer to a brand name does not provide a complete picture. We define recommendation, definition, comparison, and trust questions that customers might ask, then collect answers by model and language to identify how the brand is described and what is missing.

Deliverable	What it shows
Model-level results	Mentions, citations, visibility relative to comparison targets, questions and answers used
Answer and source evidence	Confused titles or names, missing facts, and actual cited sources
Improvement priorities	Page content, metadata, and viewing-path information to address first
Dated comparisons	Changes in results with the query, model, language, and measurement date recorded

Separate checks for brands, content, and discoverability

Brand diagnostics examine how companies and services are mentioned and cited. Content diagnostics check facts about works, people, and original stories across languages. Discoverability diagnostics examine whether content information connects to the right viewing paths.

Brand check	Information examined
Readability	Heading hierarchy, page structure, and readable content
Identity	Company and product names, aliases, and identifiers
Mentions and citations	Names and sources that actually appear in AI answers
Crawler access	Access policies and content availability
Structured data	Missing or inconsistent structured information
Trust evidence	Identifiable authors, organizations, and sources
YouTube	Visibility and connections for channel and video information
Languages	Korean and English content and language links

These eight areas are TWIGFARM’s diagnostic framework, not an AI provider’s official scoring system. Files such as llms.txt are not treated as universal AI-search requirements.

Compare models and languages

Recommendation, definition, alternatives, and trust scenarios are assessed across three AI models, comparing Korean and English answers. This can reveal a brand that appears in one model but not another, or an English title confused with a different work.

Aliases and identifiers help distinguish matching names. Comparisons with competitors use equivalent query conditions to examine relative visibility.

From content discovery to viewing paths

Check	What a content business can examine
Metadata completeness	Titles, descriptions, credits, and identifiers
Demand signals	Relative interest by work and language market
Viewing paths	Accurate platform and regional availability information
Reach	The languages and models in which content appears

FAST and broadcast operators can investigate missing viewing paths; IP owners can review title information; international teams can compare language markets. These indicators identify improvement opportunities, not measured revenue gains or losses.

Repeat measurements and interpretation

Answers can vary even for the same question, and models change over time. Record the measurement conditions and examine trends and relative comparisons. Changes to pages may take different amounts of time to appear in different AI services.

Adoption begins by agreeing on targets, comparisons, languages, and questions, then selecting actionable changes and measuring again.