What Is a Prompt?
A prompt is the input you use to ask an AI system for a response: a question, an instruction, or a task description in natural language. ChatGPT, Gemini, Perplexity, and similar systems generate their output directly from this prompt and the context it contains. For search, this shifts the rules of the game: Prompts are the new search queries, and anyone who wants to appear in AI answers must deliver content these systems pick up as a source.
Term profile
| Attribute | Details |
|---|---|
| Part of speech | Noun: the prompt, the prompts; literally “a cue to act” |
| Related practice | Prompt engineering: the deliberate crafting of prompts |
| Field | Artificial intelligence, GEO, AI visibility |
| Related terms | Generative engine optimization, LLMO, Agentic SEO |
How does a prompt work?
A prompt controls what a language model generates: The system breaks the input into tokens, calculates the most likely continuation, and formulates the response from that. The more precisely the prompt describes the task, context, and desired format, the more useful the output.
A complete prompt contains up to 4 building blocks:
- Task: What should the system do? (“Explain”, “Compare”, “Create a list”)
- Context: Background knowledge, target audience, or role (“for a trade business with 20 employees”)
- Format: Table, bullet points, text length, language
- Examples: Patterns the response should follow
Modern AI search systems add an intermediate step to the prompt: They fetch web sources matching the query and build their content into the answer, often with source citations. This is exactly where it is decided which websites get cited.
How do prompts differ from search queries?
Prompts and search queries differ in length, form, and expectation: A classic search query consists of a few keywords (“seo agency munich”), while a prompt is a full sentence or paragraph with context (“Which SEO agency suits a machine builder with 100 employees, and what should I look out for in the contract?”). The search query delivers a list of links to evaluate yourself; the prompt delivers a fully formulated answer.
For companies, this creates a measurable difference: With search queries, pages compete for positions; with prompts, they compete over whether the AI names them as a source. A user who reads an AI answer clicks less often, but is far more qualified when they do.
What do prompts mean for SEO and GEO?
Prompts change what content is optimized for: Instead of only writing for keyword rankings, websites now also write for citability in AI answers. This discipline is called generative engine optimization. Citable content offers clear definitions, well-sourced numbers, and paragraphs that stand on their own, because AI systems extract individual passages from the page.
Because prompts are longer and more specific than search queries, content that fully answers concrete detail questions wins. How to prepare your website systematically for such answers is covered on our AI visibility (GEO) service page; our article on ChatGPT SEO provides the introduction to the topic.
Frequently asked questions about prompts
What is prompt engineering?
Prompt engineering is the deliberate crafting and refining of prompts so an AI system reliably delivers the desired result. It includes role instructions, examples, and format specifications.
Can I influence whether an AI cites my website?
Yes, indirectly. AI systems prefer to cite pages with clear statements, clean structure, and machine-readable data. A mention cannot be bought directly.
Are prompts replacing classic Google search?
No, both exist side by side. Short, unambiguous queries remain faster in classic search; complex questions with context increasingly move to AI systems. Websites therefore need both: rankings and citability.
Welt der SEO lernen?
