After you have used ChatGPT or Claude for a while, you start noticing that stronger and weaker prompts have a shape to them. Once you can name the relevant parts, you can diagnose a disappointing result instead of rewriting blindly.
The five ingredients are below, with worked examples, and then how they fit together. By the end you will have a template you can adapt to important prompts in different tools, plus a validation step for checking the result.
For recurring work, think of the prompt as a task contract. It should specify the role, context, task, constraints, output format, and how uncertainty should be handled. That makes the prompt reusable and reviewable.
The five parts
In order:
- Role: what perspective or working stance should the AI use for this prompt?
- Context: who you are, the situation, what has happened so far.
- Task: what exactly you want the AI to do.
- Constraints: length, tone, things to avoid, things to include.
- Format: the shape of the output you want.
All five do not appear in every prompt. A casual question does not need them all, and examples or source material may matter more than a role. For work where the result matters, use the list as a diagnostic rather than a mandatory formula. Current vendor guidance likewise emphasizes clear instructions, relevant context, examples, output requirements, and evaluation, although neither vendor defines this exact five-part standard (OpenAI’s prompt engineering guide; Anthropic’s prompt engineering overview).
1. Role
The role is the perspective or working stance you want the model to use. It can help define the expected audience, vocabulary, and approach, but a direct task instruction may work just as well. A role does not give the model real credentials or access to professional judgment it does not have.
Weak:
Explain pension funds.
Better:
You are a clear explainer who helps mid-career employees in Estonia understand how pension schemes work. You use plain English, define jargon once, and stay at the concept level: no product picks, contribution amounts, or personal retirement plans.
Explain how pension funds work conceptually.
The model is not becoming an adviser. The instruction asks it to match the style and content density of clear explanatory text. This can make the answer more concrete without turning a literacy request into regulated advice.
A few notes on roles:
- Prefer task-relevant detail. “A careful explainer of Estonian corporate accounting who distinguishes concepts from filing advice” carries clearer boundaries than “an expert.”
- Use roles for audience, tone, and scope, not for filing or advising. A role is not a licence or proof of expertise.
- A disposition such as “skeptical reviewer,” “patient teacher,” or “careful editor” can be useful when it changes the requested behavior. Test it against equivalent concrete instructions if consistency matters.
- Avoid invented biographies and stacked status claims. “You are a world-class authority with twenty years of experience” does not create those credentials. State the required method, audience, and limits instead.
2. Context
Context is everything the model would need to know about your situation that is not obvious from the task itself. Your role, your team, the project you are working on, what has been tried, what failed, who the audience is, anything that affects what a good answer would look like.
Weak:
Draft a project update for my manager.
Better:
About me: I am a product manager at a B2B SaaS company in Tallinn. I have been leading a project to migrate our authentication system, which started two months ago.
What has happened: We hit a major issue last week with a third-party identity provider. The fix means we will deliver two weeks later than planned. The team is otherwise doing well.
My manager: senior product VP, prefers very direct updates, dislikes apologies and “we’re committed to” filler, wants to know risks and what I’m doing about them.
Draft a project update.
Notice how much the answer changes once the model knows who you are, what happened, and who is reading. It is not “more context for the sake of it.” Each line shapes the output.
Three context fields worth checking:
- Who you are. Your role, seniority, what you do.
- What has already happened. The history, the previous attempts, the prior decisions.
- Who the output is for. The audience. Their preferences, expertise, constraints.
If you run the same kind of approved prompt repeatedly, you can save stable instructions in a Custom GPT or Claude Project when your plan and workspace support that feature. Keep confidential or regulated data out unless the account, contract, and organizational policy permit it, and review saved instructions when the workflow changes.
3. Task
The task is the actual request. This is the part most people lead with. It is also the part that is often fine the way you usually write it, such as “draft an email,” “summarise this document,” or “give me three options,” provided everything else around it is doing its job.
A few useful task patterns:
- Generate variants. “Give me three drafts.” “Produce five options.” Variants give you alternatives to compare instead of asking one draft to carry the whole decision.
- Interview first, then act. “Before drafting, ask me five questions you would need answered. Wait for my answers.” Use this when missing inputs would materially change the draft.
- Critique, do not rewrite. “Find what is wrong with this. Quote specific lines. Do not rewrite.” Useful when you want the model to sharpen your work, not replace it.
- Compare and recommend. “Compare X and Y on these four dimensions. Then recommend one, and tell me what would change your answer.” Use this for low-stakes choices (wording, layout, meeting agenda). Do not use it to pick loans, medical options, or legal positions.
- Apply a framework. “Analyse this using the [STAR / SWOT / RACI / Five Whys / ICE] framework.”
The pattern is the same throughout: be specific about the verb you want the model to do.
4. Constraints
Constraints are restrictions on the output: length, tone, format, vocabulary, things to include, things to avoid.
Without relevant constraints, a model has to infer choices such as length, tone, scope, and audience. Explicit boundaries reduce that ambiguity and make the result easier to assess.
Useful constraints:
- Length. “Under 100 words.” “Three sentences max.” “One paragraph.” “Two pages.” Choose a limit that fits the use rather than assuming that more detail is better.
- Tone. “Warm but professional.” “Sharp and direct.” “Conversational, slightly self-deprecating.” “Match the tone of [paste example].”
- Vocabulary. “Do not use the words ‘leverage,’ ‘utilize,’ or ‘going forward.’” “Avoid jargon.” “Use only words a 12-year-old would understand.”
- Format-level constraints. “No preamble.” “No closing summary.” “Don’t start with ‘Great question.’”
- What to include. “Include one specific example.” “Reference the data I just gave you.”
- What to exclude. “Don’t give me an apologetic introduction.” “No ‘Great question’ opener.” For low-stakes drafting you can ask for fewer hedges; do not strip uncertainty markers or professional-advice boundaries when the topic is money, health, law, or safety.
Prefer positive, concrete instructions that describe the desired result, then add narrow exclusions where a recurring failure needs to be blocked. For example: “Start directly with the recommendation. Do not open with ‘Great question.’” Long lists of prohibitions can become harder to follow than a short statement of what good output should contain.
5. Format
Format is the shape of the output. Bullets, table, numbered list, paragraph, JSON, markdown, code block, plain text. Specifying format can reduce cleanup later.
Examples:
Return as a markdown table with three columns: option, pros, cons.
Return as a JSON object with the keys “summary,” “decisions,” and “open_questions.”
Return as three short paragraphs, no headings.
Return as a numbered list of five items, each one sentence long.
A particularly useful pattern when you want to use the output downstream (paste into a doc, feed into another prompt, post somewhere): show the model the format with a brief template:
Return in this format:
Decision: […]
Top three arguments for:
- […]
- […]
- […]
Top three arguments against:
- […]
- […]
- […]
What would change your answer: […]
The model can fill in the blanks, making the output more consistent and easier to review. If software must parse the result, use the product’s structured-output or schema feature where available and validate the returned data instead of relying on a visual template alone.
6. Validation
The sixth part is not always part of the prompt itself, but it is part of any serious workflow: how will you know the answer is good enough?
Useful validation instructions:
If you make a factual claim, mark whether it came from the source text I provided or from general background knowledge.
If any required field is missing, write
[missing]instead of guessing.
Before the final answer, list the three assumptions you made.
After drafting, give me a short checklist of what I should verify before using this.
Validation matters because a polished answer can still be wrong. A good prompt does not merely ask the model to produce output. It tells the model how to expose uncertainty, missing inputs, and review points.
Putting it together: a worked example
Let’s combine all five on a real task. Suppose you need to write a polite but firm reply to a client who is asking for free additional scope on a project.
Without the template:
Write a reply to my client who wants more work for free.
You get something generic. With the template:
Role: Help me write as a direct, friendly freelance designer setting a clear project boundary without inventing policy, contract terms, or client history.
Context: I am a freelance designer. A client (a startup founder I have worked with for a year) just asked me to add two new pages to a project that is already in its final stage. The scope was clearly fixed in our contract. The relationship is otherwise good. I do not want to lose them, but I also do not want to do the work for free.
Task: Draft three reply versions.
Constraints:
- Under 150 words each
- Warm but unambiguous: I am saying no to free work, not negotiating
- Offer a clear alternative path (e.g., a small additional engagement)
- No “I hope this finds you well”, no apologies for the boundary
- Do not include “Best regards” or any sign-off; I’ll add my own
Format: Three numbered versions, each labelled with its tone (e.g., “1. Warm and explanatory”), and a one-line “when to send this one” note under each.
The second prompt gives the model a better-defined drafting task. Review all three versions, check that none invent facts or commitments, choose the closest one, and edit it before sending.
A few patterns that combine the five
Some prompt shapes worth trying:
The reverse interview. Role + context, then a task that starts with “Ask me five questions you would need answered to do this well.” Lets the model help you sharpen the prompt itself.
The three-variant drafter. Role + context + “draft three versions” + length and tone constraints + numbered format. Useful when comparing alternatives is part of the job.
The structured analyst. Role + context + “apply this framework” + “use this exact section structure” + format. The right shape for any analytical task.
The patient tutor. Role (a tutor) + context (you, your level) + task (teach me, quiz me) + constraints (“one question at a time, wait for me”) + format (structured by phase). Combines all five and produces a useful tutoring loop.
You will develop your own. These patterns are variations of the same template: five ingredients arranged in a practical starting order, not a universal law.
A small habit that compounds
After any meh AI answer, pause for two seconds. Ask: which of the five did I leave out?
- Did the model not know who it was being? → role.
- Did it not have my situation? → context.
- Did it misunderstand the verb? → task.
- Was the output too long, too formal, too padded? → constraints.
- Was the shape of the answer the wrong shape? → format.
You may find a missing ingredient, an unclear instruction, or a weak example. Repair that specific gap and try again. This habit teaches you from your own prompts and outcomes rather than from a generic formula alone.
Five parts plus validation. Use them deliberately until reaching for them is automatic.
