Large language model (LLM)
DefinitionAI
Large language model (LLM)
A model trained on vast text corpora to predict language, which turns out to cover writing, summarising, reasoning and conversation - GPT, Claude and Gemini are LLMs. For marketers they are simultaneously a production tool, a research assistant, and a new gatekeeper answering customers' questions about the category.
Evidence from the library
LLMs act as systematic, product-centric deciders at the evaluation stage of the customer journey.Principle~66% / ~25% / ~10%Brand equity accounts for about two thirds of LLM visibility, current marketing activity about a quarter, and citation volume only about a tenth.StatHuman ad perception works in wide beam (gestalt, story, emotion); LLM perception works in narrow beam (features, claims, structure).PrincipleAd translations get cross-checked across multiple LLMs instead of run through just one, trading speed for consistency.Principle63%Long-term brand equity drives 63% of LLM visibility.Stat
Source
Marketing Definitions
standard AI terminology
More from this source
We wrote about this
The End of the Media Buyer and the Rise of the AI Agent-in-ChiefAgentic AI and Safe Pixels: The Brand Strategy for the New Creative PipelineTranslation Is Not Localization (Ask ElevenLabs' Korean Ads)Your AI campaign looks like a campaign. That is the problem.AI slop is not bad content. It is expected content.Synthetic Actors and Search Agents: The Brand Strategy for an Automated Web