ChatGPT Prompts: Patterns and Examples That Get Better Answers

ChatGPT responds dramatically better to structured prompts than to bare questions. These patterns are reusable across writing, coding and analysis tasks.

The persona pattern

You are a [specific expert]. Your audience is [specific reader]. [Task]. Constraints: [limits]. Format: [structure].

Specificity is the entire trick. “You are a copywriter” is weak; “You are a direct-response copywriter who writes for skeptical technical audiences” measurably changes tone and word choice.

The critique-then-revise pattern

Ask for evaluation before generation: “First list three weaknesses of the draft below, then rewrite it fixing all three.” This consistently beats “improve this text,” because the model commits to concrete problems before revising.

The example-driven pattern

When format matters, show one example input and output, then provide the real input. One good example outperforms three paragraphs of format description, especially for tables, JSON and structured summaries.

The step-constrained pattern

For multi-part work, number the steps and require numbered output: “Complete these 4 steps. Label each answer 1-4.” Unnumbered multi-part prompts lose steps silently.

Common ChatGPT prompting mistakes

• Asking for everything at once. Long unstructured wish lists produce shallow coverage of each item.
• No length constraint. ChatGPT defaults to verbose; state a word or item limit.
• Ignoring the context window. Very long conversations degrade; start fresh threads for new tasks and check size with a token counter.
• Pasting Markdown everywhere. ChatGPT answers arrive formatted; strip it with a Markdown cleaner before using the text in emails or documents.

A complete example

You are a hiring manager at a mid-size SaaS company. Rewrite the job description between the quotes for a senior backend engineer role. Audience: experienced engineers skeptical of buzzwords. Keep it under 250 words, plain language, and end with 4 bullet points of concrete responsibilities.

Every pattern above appears in that prompt: persona, audience, task, constraints and format. That is the whole method.