Retrieval-augmented generation (RAG)
Generating an answer with information retrieved from external sources.
What is Retrieval-augmented generation (RAG)?
RAG systems retrieve relevant documents or passages, then provide that context to a language model. Retrieval can improve freshness and grounding, but answer quality still depends on source selection, context limits, and the model's use of evidence.
An assistant searches a help center before answering a product support question.
Why it matters for AI visibility
AI answers can shape awareness and buying decisions before a person visits a website. Understanding retrieval-augmented generation (rag) helps teams connect what a model says with the prompts, sources, and technical signals that influenced the result. The useful next step is to observe the evidence, compare it over time, and act on the specific gap.
How to measure or use it
- Define the scope. Choose the relevant market, prompts, models, locations, and time window.
- Keep the evidence. Save the complete answer, its citations, and the page or claim being evaluated.
- Compare patterns. Look for repeated movement across prompts and models instead of reacting to one response.
- Take a specific action. Improve the owned page, correct a claim, earn third-party inclusion, or create better evidence.