# Role, Audience and Tone — Prompt Engineering

Source: https://www.geekswithgeeks.com/en/prompt-engineering/c-role

> Use personas and audience descriptions to steer style and depth.

## Who is speaking, and to whom

Naming a **role** ("senior Python reviewer", "patient teacher for Class 10 students") shifts vocabulary, depth and priorities. Naming the **audience** matters at least as much: the same explanation of recursion differs for a beginner and an expert. Describe **tone** with concrete words ("friendly, no jargon, short sentences") rather than single adjectives. A role does not give the model knowledge it lacks, so it will not make a weak answer correct, and it is not a substitute for facts or constraints. Keep personas honest and relevant to the task.

## Same question, different audience

Only the audience line changes between these two prompts.

```text
Explain what an API is.
Audience: a 14-year-old who has never coded. Use a restaurant analogy. 4 sentences.

Explain what an API is.
Audience: a backend engineer. Mention contracts, versioning and idempotency. 4 sentences.
```

## Roles do not add knowledge

"Act as a doctor" changes tone, not medical accuracy. Provide the facts, and keep a human in the loop for high-stakes advice.

**Quiz:** What does naming the audience change?

- [x] Depth, vocabulary and examples
- [ ] The model size
- [ ] The tokenizer
- [ ] The price per token

*Answer:* Depth, vocabulary and examples. The same facts are presented differently for different readers.