Prompt Engineering: The Skill That Multiplies AI Value

Why does asking the right question matter so much?

Core prompt engineering techniques

Core prompt engineering techniques

Feature	How it works	Best for	Example
Zero-shot	Give the model a task with no examples	Simple, well-defined tasks the model already understands	Summarise this meeting transcript in 5 bullet points
Few-shot	Provide 2-3 examples of the desired output before asking	Tasks where format or style matters, or the model needs calibration	Here are 3 example customer replies we liked. Now write one for this complaint.
Chain-of-thought	Ask the model to reason step by step before answering	Complex analysis, calculations, or multi-step decisions	Think through the pros and cons of each vendor, then recommend one.
System prompts	Set the AI's role, rules, and constraints before the user interacts	Custom AI applications and agents that need consistent behaviour	You are a financial analyst. Only use data from Q3 reports. Never speculate.
Iterative refinement	Build on previous outputs — refine, expand, or redirect	When the first output is close but needs adjustment	Good start. Make the tone more formal and add a data table.

💡 Exam tip: Know when to use each technique

The exam tests your ability to match a technique to a scenario:

• Zero-shot works when the task is straightforward and the model understands it natively. No extra context needed.
• Few-shot is essential when you need a specific format, tone, or structure that the model wouldn’t produce by default.
• Chain-of-thought is critical when the answer requires reasoning — budgets, comparisons, risk assessments.
• System prompts are how organisations standardise AI behaviour across users and applications.
• Iterative refinement is the most practical technique — most real-world prompting involves a conversation, not a single shot.

Five elements of an effective prompt

Every great prompt includes some combination of these five elements:

Element	What it does	Example
Role	Tells the AI who it should behave as	”You are a senior procurement analyst”
Task	States exactly what you want done	”Compare these three vendor proposals”
Context	Provides background the AI needs	”We’re a manufacturing company with 5,000 staff and a $2M annual IT budget”
Format	Specifies the output structure	”Present as a table with columns for cost, capability, and risk”
Constraints	Sets boundaries and rules	”Keep the response under 300 words. Don’t recommend any vendor we’ve used before.”

You don’t need all five every time. But the more elements you include, the more precise and useful the output becomes.

Before and after: The prompt difference

Here’s the same task with a weak prompt versus a strong prompt:

ℹ️ Before: Weak prompt

Prompt: “Write about our Q3 results.”

Result: A generic, vague summary that could apply to any company. Missing key numbers. Wrong tone for the audience. Requires 20 minutes of rewriting.

ℹ️ After: Strong prompt with all five elements

Prompt: “You are the CFO of a mid-sized consulting firm (Role). Write a Q3 results summary for the board (Task). Revenue grew 12% to $47M, margins improved from 18% to 22%, and we added 3 enterprise clients (Context). Use a professional, concise tone with bullet points for key metrics followed by a 2-sentence outlook (Format). Keep it under 250 words and don’t include competitor comparisons (Constraints).”

Result: A polished, board-ready summary that needs minimal editing. The specific numbers are included. The format matches expectations. Done in 2 minutes instead of 20.

Real-world scenario: Elena’s prompt playbook

👔 Elena, CEO of Meridian Consulting, realises that her 200 consultants get wildly different results from Copilot. Some love it. Others say it’s useless. The difference? How they ask.

Elena commissions a prompt playbook — a library of tested, optimised prompts for common consultant tasks:

• Client proposal drafts — Few-shot prompts with 3 examples of winning proposals
• Meeting summary emails — Zero-shot prompts with format constraints (bullet points, action items, owners)
• Research analysis — Chain-of-thought prompts that walk through methodology before conclusions
• Executive briefings — System prompts that enforce Meridian’s house style and confidentiality rules

After rolling out the playbook, consultant satisfaction with Copilot jumps from 45% to 82%. Average time saved per consultant goes from 2 hours per week to 6 hours per week.

💡 Business impact: Prompt quality is an organisational capability

Elena’s experience highlights a critical insight for leaders: prompt engineering is not an individual skill — it’s an organisational capability. The companies that get the most value from AI invest in:

• Shared prompt libraries that codify best practices
• Training that teaches employees how to iterate on prompts
• Governance that ensures prompts don’t accidentally expose sensitive data or bypass policies

The exam expects you to understand that prompt quality directly impacts AI ROI at the organisational level.

Key flashcards

Question

What is the difference between zero-shot and few-shot prompting?

Click or press Enter to reveal answer

Answer

Zero-shot gives the model a task with no examples — it relies on the model's built-in knowledge. Few-shot provides 2-3 examples of the desired output so the model can match the pattern, format, or style.

Click to flip back

Question

When should you use chain-of-thought prompting?

Click or press Enter to reveal answer

Answer

When the task requires multi-step reasoning — budgets, comparisons, risk assessments, or any scenario where the AI needs to 'think through' the problem before giving an answer. It reduces errors on complex tasks.

Click to flip back

Question

What are the five elements of an effective prompt?

Click or press Enter to reveal answer

Answer

Role (who the AI should be), Task (what to do), Context (background information), Format (output structure), and Constraints (boundaries and rules).

Click to flip back

Question

Why is prompt engineering an organisational capability, not just an individual skill?

Click or press Enter to reveal answer

Answer

Because consistent AI value requires shared prompt libraries, training programmes, and governance — not just individual experimentation. Organisations that standardise prompt quality see 2-3x more value from AI investments.

Click to flip back

Knowledge check

Knowledge Check

Elena's team creates a prompt that includes three examples of well-written client proposals before asking Copilot to draft a new one. Which technique is this?

AZero-shot promptingBFew-shot promptingCChain-of-thought promptingDSystem prompting

Check Answer

Knowledge Check

Priya, Meridian Consulting's CFO, asks Copilot: 'Analyse our Q3 budget. Walk through each department's spending step by step, then identify the top 3 areas of overspend.' Which prompting technique is Priya using?

ASystem promptingBIterative refinementCChain-of-thought promptingDZero-shot prompting

Check Answer

Next up: RAG and Grounding: Making AI Use YOUR Data — how to connect AI to your organisation’s knowledge and reduce fabrications.