Writing clear prompts is the key to getting useful AI responses that match your firm's needs. This guide walks you through the five essential components of an effective prompt—from setting the AI's role to adding helpful examples. By the end, you'll understand how to structure prompts that produce relevant, accurate outputs tailored to your professional services context. We've included practical examples from talent development scenarios to show you exactly how it works.
The five components of an effective prompt
Effective prompts typically follow this sequence:
State the role/perspective the AI should take
Define the specific task or output needed
Provide relevant context
Specify constraints and format
Add examples if helpful
Every strong AI prompt includes these five elements working together. Think of them as building blocks—each one adds clarity and precision to ensure the AI understands exactly what you need.
1. Set the role or perspective
Start by telling the AI what perspective to take. This shapes how it interprets your request and the tone it uses.
Why it matters: A vague role like "act as an expert" doesn't give enough direction. The AI needs to know what kind of expert and for which audience.
How to do it well: Be specific about the role and context. Instead of "you are an adviser", try "you are a talent development adviser in a global law firm" or "you are an executive coach specialising in professional services".
Example: "You are Compass, an objective-setting adviser for professional services" or "You are an expert Executive Coach and Talent Specialist within a top-tier global law firm".
2. Define the task or output
Clearly state what you want the AI to produce. Be explicit about the type of output—whether it's a summary, analysis, recommendations, draft text, or structured feedback.
Why it matters: Vague requests like "help me with this" lead to generic responses. If you don't specify the output type, the AI will guess—often incorrectly.
How to do it well: Use precise verbs like "summarise", "analyse", "draft", "identify", or "suggest". Specify quantity where relevant (e.g. "3–5 key strengths").
Example: "Analyse feedback you've received from colleagues. Provide a personal, encouraging summary to support your professional development" or "Produce a summary based on the four domains".
3. Provide relevant context
Give the AI the background information it needs to understand your situation. This includes your audience, purpose, domain-specific knowledge, and any relevant frameworks or documents.
Why it matters: Too much context can overwhelm; too little leaves the AI guessing. Irrelevant context can lead to outputs that miss the mark entirely.
How to do it well: Include who the output is for, why it matters, and what it relates to. Reference specific frameworks, competencies, or strategic plans when applicable. Link context to your firm's terminology and values.
Example: "The capability framework shows role expectations. Strategic plans include specific priorities, targets and metrics. Ground your suggestions in these—reference actual plan priorities and numbers rather than generic themes" or "Use the provided document as context for partner expectations".
4. Specify constraints and format
Define the boundaries for your output: tone, style, length, structure, and what to avoid. This ensures the response fits your needs and context.
Why it matters: Without constraints, outputs can be too long, use the wrong tone, or include unwanted content. Overly restrictive constraints can stifle useful insights, so find the right balance.
How to do it well: Be clear about tone (formal, conversational, encouraging), length limits (word count, bullet points), structural requirements (headings, sections, format), and exclusions (what not to include). Specify language preferences (e.g. UK English).
Example: "Keep it encouraging and constructive. Bullet points for readability. Keep total response under 150 words. Use UK spelling" or "Use only the text provided below. Do not invent facts. If evidence is missing or unclear, flag it explicitly".
5. Add examples (when helpful)
Where appropriate, provide examples of good outputs or reference existing content that matches your desired style. This helps the AI understand exactly what "good" looks like in your context. However note that AI will use these examples when it doesn't have sufficient data. Too many or poor examples will also constrain output - use with caution and test.
Why it matters: Examples that don't align with your actual needs can mislead the AI. Too many examples can constrain creativity when you want variation.
How to do it well: Provide 1–2 strong examples that demonstrate tone, structure, or content style. Use examples to illustrate how to apply your frameworks, not just what to say. Examples work especially well for format and tone guidance.
Example: Including a sample output format like "Start each bullet with a brief heading (in bold font) followed by a colon and then the detail" or showing how to reference competencies: "e.g. 'Increase BD meetings with target clients from 2 to 6 monthly' not 'Enhance business development'".
Start simple and refine as you go. You don't need to perfect every prompt from the start. Try it, review the output, then adjust your prompt based on what worked and what didn't. Testing is key!
The examples below review 3 default prompts from Performance Leader against the guidance above and suggest how you might adapt them to your context.
Example 1: Objective setting assistant
You are Compass, an objective-setting advisor for professional services.
STYLE: Advisor, not interviewer. Offer specific, measurable suggestions (e.g. "Increase BD meetings with target clients from 2 to 6 monthly" not "Enhance business development"). UK spelling. Use bullet points and bold for clarity, but no headings.
FOR SENIOR ROLES: Draw objective ideas from the strategic plan priorities provided rather than generic categories. Use specific targets and metrics from the plan (e.g. revenue goals, growth percentages, market focus areas) to make suggestions concrete and measurable.
CONTEXT: The capability framework shows role expectations. Strategic plans include specific priorities, targets and metrics. Ground your suggestions in these — reference actual plan priorities and numbers rather than generic themes. If the user is unsure what to focus on, point them to specific priorities from their chosen plan or capabilities from their framework.
CONVERSATION: Keep the dialogue natural. Don't always end with a question - sometimes let your advice stand on its own. When you do ask, vary the phrasing. AVOID: Starting with validation, restating what user said, generic advice, or repetitive question formats.
How this prompt applies the guidance
Role: "You are Compass, an objective-setting advisor for professional services" — establishes the specific perspective and expertise level the AI should adopt.
Task: Implicit throughout — to provide specific, measurable objective suggestions rather than generic advice.
Context: References capability frameworks and strategic plans as grounding documents. Explicitly states that senior roles should draw from strategic plan priorities with specific targets and metrics.
Constraints: Detailed style requirements (UK spelling, bullet points, no headings), conversational tone guidance (vary phrasing, don't always end with questions), and clear exclusions (avoid validation, generic advice, repetitive formats).
Examples: Provides concrete examples of good vs. poor suggestions ("Increase BD meetings with target clients from 2 to 6 monthly" vs. "Enhance business development").
What you might adapt
Role specificity: Change "Compass" to your own tool name or adjust the professional services context to match your industry (e.g., "accounting firm", "consultancy").
Context documents: Replace references to "capability framework" and "strategic plan" with your organisation's actual performance management tools or planning documents.
Tone and style: Adjust the conversational style guidance based on your firm's culture — some organisations may prefer more formal, structured responses rather than natural dialogue.
Examples: Substitute the business development example with objectives more relevant to your sector or the specific roles you're supporting (e.g., client retention metrics, project delivery targets, technical certifications).
Sample output below:
Example 2: Producing a summary of feedback received
Analyse feedback you've received from colleagues. Provide a personal, encouraging summary to support your professional development.
**💪 YOUR KEY STRENGTHS:**
Summarise 3-4 recurring strengths your colleagues highlight. Use direct language (e.g., "Your colleagues consistently praise your...").
**🎯 AREAS TO DEVELOP:**
Identify 2-3 development opportunities mentioned in feedback. Frame constructively as growth areas.
**🔄 RECURRING THEMES:**
Note any patterns or common threads across multiple pieces of feedback.
**💡 ACTIONABLE NEXT STEPS:**
Suggest 2-3 specific actions you could take based on this feedback.
**Format Requirements:**
- Use emojis for visual structure
- Address the person directly using "you" and "your"
- Keep it encouraging and constructive
- Bullet points for readability
- Keep total response under 150 words
- Use UK spelling (e.g., "summarise", "recognise")
**Tone:**
- Positive and motivating, even when discussing development areas
- Frame feedback as opportunities for growth
- Celebrate strengths genuinely
JSON Feedback:
How this prompt applies the guidance
Role: Establishes the AI as an analyst focused on feedback interpretation, with an encouraging, developmental perspective.
Task: Clearly defined — analyse feedback and provide a structured summary with four specific sections (strengths, development areas, themes, next steps).
Context: Implicit context about professional development and colleague feedback. The structured output format itself provides context about how feedback should be organised.
Constraints: Highly specific formatting requirements (emojis, direct address, bullet points, 150-word limit, UK spelling) and detailed tone guidance (positive, motivating, growth-focused).
Examples: The format itself serves as an example, showing exactly how to structure the output with emoji headings and direct language patterns.
What you might adapt
Output structure: Modify the four sections to match your firm's feedback framework or performance review categories (e.g., add sections for technical skills, client feedback, or leadership behaviours).
Length constraints: Adjust the 150-word limit based on your needs — longer summaries for comprehensive annual reviews, shorter for quarterly check-ins.
Visual elements: Change or remove emoji usage to match your organisation's communication style — some firms prefer more formal, emoji-free formats.
Tone calibration: Adjust how constructively development areas are framed. Some contexts may require more direct language about performance gaps, whilst others benefit from the softer "growth opportunity" framing.
Sample output below:
Example 3: In Review feedback summary
Role: You are an expert Executive Coach and Talent Specialist within a top-tier global law firm, and are assisting an employee (non-partner) to produce a concise, evidence-based self-assessment summary for an annual review.
Objective: Analyse the user's self-assessment inputs. Provide constructive feedback to the employee to help them strengthen their review before they submit it to their manager.
Strategic context: The business has launched a new Excellence Framework and simplified the performance review process. A core focus is strong goals, improving collaboration and encouraging honest constructive feedback.
Instructions:
Executive summary: Reflect on the overall quality of the submission. Highlight strengths (clear, specific, honest articulation of contribution and challenges) and weaknesses (vague, superficial, weak comments)
Review "My contribution" section: Review inputs in this section and provide feedback. Consider our Excellence Framework (provided below).
Review "Looking forward" section: Review inputs in this section and provide feedback. Consider our Excellence Framework (listed below).
Review against Excellence Framework: Provide a short summary of how the inputs have mapped against the Excellence Framework.
Constraints:
- Use only the text provided below. Do not invent facts. If evidence is missing or unclear, flag it explicitly.
- Do not write the review for them. Provide a "Feedback Report" with suggestions.
- Ensure it is formatted so it's easy to read. Don't bullet headings.
- Keep the tone objective, encouraging and professional. Please use UK English.
How this prompt applies the guidance
Role: "You are an expert Executive Coach and Talent Specialist within a top-tier global law firm" — highly specific role with clear domain expertise and organisational context.
Task: Multi-part objective clearly defined — analyse self-assessment inputs, provide constructive feedback, and review against a specific framework. Explicitly states what the AI should not do (write the review for them).
Context: Rich strategic context provided (new Excellence Framework, simplified review process, core focus areas). References specific documents and frameworks that ground the analysis.
Constraints: Evidence-based requirement ("Use only the text provided below. Do not invent facts"), format specifications (easy to read, no bullet headings), tone guidance (objective, encouraging, professional, UK English), and clear output type (Feedback Report with suggestions, not completed review).
Examples: Whilst no explicit examples are shown, the structured instruction list implicitly demonstrates the expected analytical approach and output sections.
What you might adapt
Role and industry: Change "top-tier global law firm" to your organisation type and adjust the expertise level to match your context (e.g., "HR Business Partner in a mid-sized consultancy").
Framework references: Replace "Excellence Framework" with your organisation's actual competency framework, values framework, or performance criteria. List the specific domains or competencies if they're standard across your firm.
Strategic context: Update the business context to reflect your current organisational priorities — this helps the AI understand what "good" looks like in your specific environment right now.
Evidence constraints: Adjust how strictly the AI should stick to provided text. For some use cases, you might want the AI to draw on general professional development knowledge rather than being limited to submitted content only.
Sample output below:
