Claude Prompt Anatomy: 10 Steps to Better AI Answers
This 10-step AI prompt framework helps produce accurate, useful, well-structured content and research. It combines planning, role definition, task instructions, context, formatting, constraints, examples, success criteria, clarification, and factual verification.
1. Think and Plan
Understand the objective, identify the audience and search intent, plan the work, identify research requirements, and resolve important dependencies before drafting.
- Identify the primary objective and desired outcome.
- Determine the intended audience and scope.
- Identify research requirements and missing information.
- Plan the steps needed to complete the task.
- Define how the final result will be evaluated.
2. Who — Define the Role
Adopt the relevant professional role and domain expertise. Specify the perspective required for the task without claiming credentials or experience that cannot be substantiated.
Example: You are a technical SEO specialist experienced in content architecture, search intent analysis, and website optimization.
3. What — Define the Task
Define the exact deliverable, scope, intended audience, and desired outcome. Replace vague instructions with specific actions.
Example: Research the target keyword, identify related search intents, develop a content structure, and write a publication-ready article.
4. Context — Supply the Evidence
Provide the information needed to complete the task accurately. Include relevant source material, standards, dates, geographical scope, and audience requirements.
- Use supplied documents and data where relevant.
- Identify the applicable country, market, or industry.
- Use reliable research sources when external evidence is required.
- Identify missing information and conflicting evidence.
- Do not assume access to files, websites, or accounts that are unavailable.
5. Format — Define the Deliverable
Specify the output structure, headings, tables, lists, tone, length, and delivery format.
Possible deliverables include:
- SEO articles with H2, H3, and H4 headings.
- WordPress-ready HTML using only the required body content.
- Tables, reports, checklists, and comparison guides.
- Spreadsheets with formulas and documented assumptions.
- Research summaries with citations and source links.
6. Limits — Set Boundaries
Define the restrictions that the work must follow. Specify exclusions, source requirements, word limits, assumptions, and prohibited actions.
- Do not invent statistics, citations, quotations, or sources.
- Do not change source data without authorization.
- Do not present estimates as verified facts.
- Follow the requested scope and formatting requirements.
- Identify limitations when evidence is incomplete.
7. Examples — Show the Quality Standard
Provide examples of preferred and unsuitable results. Explain which characteristics should be reproduced and which should be avoided.
| Example type | Purpose |
|---|---|
| Preferred example | Demonstrates the required structure, detail, tone, and quality. |
| Unsuitable example | Shows weaknesses such as repetition, vague claims, or missing evidence. |
| Explanation | Identifies the specific features to reproduce or avoid. |
Use examples as quality references, not as permission to copy protected material or reproduce factual errors.
8. Success Criteria — Define Acceptance Tests
Set measurable standards for completeness, relevance, clarity, accuracy, originality, usability, and technical correctness.
- Every required section is present.
- Important claims are supported by appropriate evidence.
- Calculations can be checked and reproduced.
- Formatting and technical requirements are satisfied.
- The deliverable addresses the intended audience and objective.
- SEO content satisfies search intent without keyword stuffing.
9. Ask First — Resolve Uncertainty
Ask focused questions when missing information could materially change the result. If clarification is unnecessary, state reasonable assumptions and proceed.
Instruction: Before starting, identify essential missing information, conflicting requirements, and unclear assumptions. Ask only the questions that materially affect the outcome.
10. Verify Statistics and Figures — Fact-Check
Before delivering the final result, check material factual claims, statistics, dates, calculations, quotations, and citations against reliable sources. Prefer primary sources and current evidence whenever appropriate.
Fact-Checking Checklist
- Statistics: Confirm the figures against reliable sources and verify the relevant reporting period.
- Calculations: Recalculate totals, percentages, averages, conversions, and formulas independently.
- Dates: Check publication dates, update dates, and time-sensitive claims.
- Citations: Confirm that cited pages exist and actually support the associated claims.
- Technical facts: Verify specifications, software features, compatibility, and implementation details.
- SEO claims: Distinguish documented search-engine guidance from opinions, hypotheses, and unverified ranking claims.
- Uncertainty: Label estimates, distinguish facts from hypotheses, and qualify or remove unsupported statements.
- Transparency: Never claim that a fact, source, calculation, or statistic was verified if the check was not performed.
Final Delivery Rules
Deliver the requested output only after completing the applicable steps. The amount of research and verification should match the complexity and importance of the task.
For SEO articles:
- Satisfy the primary search intent early in the article.
- Use a logical heading hierarchy and natural semantic coverage.
- Include relevant entities, related questions, and useful examples.
- Use tables and lists when they improve understanding.
- Prioritize original value, accuracy, and people-first content.
- Use reliable sources for material factual claims.
- Avoid keyword stuffing, unnecessary repetition, and unsupported claims.
- Do not guarantee rankings or treat a prompt framework as a guarantee of search performance.
Conclusion
This 10-step framework separates planning, execution, clarification, and final verification. It helps create work that is easier to review, test, and use.
The essential principle is simple: plan the task, provide the right evidence, define the expected result, and verify the finished work. A well-structured prompt can improve the process, but accuracy still depends on reliable evidence, appropriate checks, and sound judgment.
