Why use iteration in prompting?

Iteration improves output quality step by step. Instead of trying to write the perfect prompt upfront, you refine the AI output through follow-up instructions — making it shorter, changing the tone, or adding specific details.

Advanced Prompting | Expert AI Prompt Techniques

Section 1

6 core advanced methods

Each method solves a specific problem that basic prompting can't. Tap any card to see a full explanation and a copyable example prompt.

🎭

Persona prompting

Assign the AI a role — shapes everything that follows

Foundational ▼

Persona prompting uses "Act as a [role]" at the start of the prompt to assign the AI a specific identity. This changes how it frames knowledge, what vocabulary it uses, and how it structures the output. A single sentence of role assignment can transform a generic answer into expert-level output.

When to use it

Any time you need domain-specific knowledge or a particular perspective — marketing copy, technical writing, coaching advice, legal summaries, or creative work.

Example prompt

Act as a senior UX designer with 10 years of experience in SaaS products. Review the following onboarding flow and identify the top 3 friction points. For each one, suggest a specific improvement with a brief rationale. [Paste your onboarding steps here]

📸

Few-shot prompting

Teach the pattern before asking for the output

High impact ▼

Few-shot prompting gives the AI one or more examples of the exact output you want before asking it to generate the real thing. The AI learns the pattern — tone, format, length, structure — from your examples and replicates it precisely. This is the fastest way to enforce a consistent output style.

When to use it

When you need a specific format that's hard to describe in words — email templates, social captions, product descriptions, headlines, or any repeatable content format.

Example prompt

Here are two examples of the LinkedIn post style I want: Example 1: "Most people write emails that get ignored. Here's the one change that fixed mine: I stopped opening with 'I hope this finds you well.' I started with the result. Try it." Example 2: "The best marketing advice I ever got: Talk about the customer's problem, not your product's features. Nobody buys features. They buy outcomes." Now write 3 LinkedIn posts in the same style about the topic of AI productivity tools.

🔗

Chain-of-thought prompting

Force the AI to reason before answering

Accuracy ▼

Adding "think through this step by step" or "reason through this before answering" to any complex prompt dramatically improves accuracy. The AI externalises its reasoning, which catches errors and produces more thoughtful conclusions — especially for analytical, logical, or multi-factor decisions.

When to use it

Analysis, strategy, problem-solving, risk assessment, comparisons — any task where the quality of reasoning matters as much as the final answer.

Example prompt

I'm deciding whether to launch a paid newsletter or a free one with sponsorships. Think through this decision step by step, considering: 1. Revenue predictability 2. Audience growth speed 3. Long-term brand positioning 4. Operational complexity After reasoning through each factor, give a clear recommendation with your rationale.

🔄

Iterative refinement

Improve output in rounds rather than one shot

Workflow ▼

Iterative refinement treats the AI conversation as a drafting process. You get a first output, identify what needs improving, and instruct the AI to fix those specific things. This is faster and more effective than trying to write the perfect prompt upfront — and it produces results that genuinely match your vision.

When to use it

Complex writing, content creation, strategy documents, code — anything where the first draft is a starting point, not the final result.

Refinement command examples

"Make the opening paragraph more direct. Cut the first two sentences." "Rewrite the third section with a more confident tone." "Add a concrete example after the second point." "Shorten by 30% without losing the key argument." "Change the CTA at the end to focus on urgency, not features."

🚧

Constraint prompting

Limits that force focus and raise quality

Quality ▼

Constraint prompting adds explicit limits to force the AI to be more disciplined — word counts, formatting rules, forbidden phrases, required inclusions. Counter-intuitively, tighter constraints produce better outputs because they eliminate filler and force every sentence to earn its place.

When to use it

Any time you need concise, punchy output — headlines, ad copy, email subjects, pitches, summaries, or any content with a hard length requirement.

Example prompt

Write a homepage headline for a project management tool targeting remote teams. Constraints: - Under 8 words - Must include a benefit, not a feature - No jargon (no "synergy", "streamline", "empower") - Must create curiosity or tension - Give 5 options

🔀

Multi-step prompting

Break complex tasks into connected steps

Scale ▼

Multi-step prompting breaks a large or complex task into sequential prompts, where each step builds on the previous output. This overcomes the limitations of a single prompt, keeps the AI focused, and produces higher-quality results than asking for everything at once.

When to use it

Content creation pipelines, research projects, product launches, course creation — any task with 3 or more distinct phases.

Example sequence

Step 1: "Research the top 5 pain points of freelance designers when managing client projects." Step 2: "Based on those pain points, create an outline for a blog post targeting freelance designers." Step 3: "Write the full post using the outline. 800 words. Friendly tone." Step 4: "Rewrite the headline 5 ways to improve click-through rate." Step 5: "Write 3 social media captions to promote this article on LinkedIn."

💡

Pro tip: You rarely need to choose just one technique. The most effective advanced prompts combine multiple methods — a persona, a few-shot example, a constraint, and a step-by-step instruction in a single prompt.

Section 2

Technique comparison

Which method to reach for, when — and what level of experience it requires.

Scroll to see full table on mobile

Technique	Best for	Level	Key phrase to use
Persona prompting	Domain-specific expertise, tone shaping	Beginner	"Act as a [role]…"
Few-shot prompting	Enforcing a specific format or style	Intermediate	"Here are two examples… now do the same for…"
Chain-of-thought	Analysis, decisions, complex reasoning	Intermediate	"Think through this step by step…"
Iterative refinement	Long-form content, writing quality	Intermediate	"Rewrite the [section] to be more [quality]…"
Constraint prompting	Concise copy, headlines, tight formats	Beginner	"Under [X] words. No [word]. Must include [Y]."
Multi-step prompting	Content pipelines, large projects	Advanced	"Step 1:… Step 2:… Step 3:…"

Section 3

Full 5-step workflow example

A complete content launch workflow using multi-step prompting. Each step builds on the last — run them in sequence in a single conversation or across separate sessions.

1

Research the topic

Use the AI to identify key pain points, questions, and angles before writing a single word.

Prompt

Act as a market researcher. Identify the top 5 pain points of small business owners when it comes to managing their online reputation. Include what questions they search for and what outcomes they want.

2

Build the outline

Convert the research into a structured content plan with clear sections and a logical flow.

Prompt

Based on those pain points, create a detailed outline for a 1200-word blog post targeting small business owners. Include H2 headings, a hook, and a CTA at the end. Focus on practical, actionable advice.

3

Draft the content

Write the full piece using the outline. This is where persona and constraint prompting combine for best results.

Prompt

Act as a senior content strategist. Write the full blog post using the outline above. Tone: professional but conversational. Avoid jargon. Include one real-world example for each main section.

4

Optimise for SEO

Refine the draft with search intent in mind — without sacrificing readability.

Prompt

Review this draft for SEO. Suggest improvements to the title tag, meta description, H2 headings, and internal link opportunities. Do not change the tone or structure — only optimise for search.

5

Repurpose for social media

Extract maximum value from the content by adapting it for different platforms and formats.

Prompt

Using the completed blog post, create: 3 LinkedIn posts (professional tone), 5 tweet-length takes (punchy, direct), and 1 email newsletter intro (150 words, warm and personal).

⚡

Time saving: This 5-step workflow replaces what would take 4–6 hours manually. Run each step in the same conversation thread so the AI retains context across all steps.

Section 4

Upgrade any basic prompt

See how layering advanced techniques transforms a simple request into a high-quality output instruction. Each upgrade adds one technique.

❌ Basic

Write a launch plan for my SaaS product.

✓ + Persona

Act as a senior growth strategist. Write a launch plan for my SaaS product.

✓ Previous version

Act as a senior growth strategist. Write a launch plan for my SaaS product.

✓ + Chain-of-thought

Act as a senior growth strategist. Think through the key phases of a SaaS launch step by step before writing. Then produce a 30-day launch plan with weekly milestones.

✓ Previous version

...Then produce a 30-day launch plan with weekly milestones.

✓ + Constraints

...Then produce a 30-day launch plan with weekly milestones. Format as a table with columns: Week, Focus area, Key actions, Success metric. No generic advice — every action must be specific and executable.

The final prompt combining all three upgrades produces a structured, expert-level launch plan — not a generic list of ideas. That's the compounding effect of layering advanced techniques.

Section 5 — Try it now

A ready-to-use advanced prompt

This prompt combines persona + chain-of-thought + constraints + multi-step. Copy it into any AI and observe how the layered techniques produce a precision output.

Act as a senior growth strategist with experience launching B2B SaaS products.

Think through this step by step before writing:
— What are the critical phases of a 30-day SaaS launch?
— What metrics matter most in each phase?
— What are the most common mistakes teams make at launch?

Then produce a 30-day launch plan with weekly milestones.

Format: a table with columns — Week | Focus area | Key actions (max 3) | Success metric
Constraints: No generic advice. Every action must be specific, executable, and measurable.

Techniques used: persona prompting + chain-of-thought + multi-step + constraints. All in one prompt.

FAQ

What is few-shot prompting?▼

Few-shot prompting means giving the AI one or more examples of the output you want before asking it to produce the real thing. The AI learns the pattern from your examples — tone, format, length, structure — and applies it to the actual task. One good example is usually enough.

Why use iteration instead of one long prompt?▼

Iteration lets you correct as you go rather than trying to anticipate every requirement upfront. A first draft shows you what's missing or wrong — follow-up instructions then fix exactly those things. This produces better results faster than spending 20 minutes writing one "perfect" prompt.

What is chain-of-thought prompting?▼

Chain-of-thought prompting asks the AI to reason through a problem step by step before giving the final answer. Adding "think through this step by step" to any complex prompt significantly improves accuracy — especially for analysis, decisions, and multi-factor problems.

When should I use multi-step workflows?▼

Use multi-step workflows for complex tasks that would overwhelm a single prompt — content pipelines, research projects, product launches, or anything requiring more than one type of output. Break the task at natural phase boundaries and run each step in the same conversation so context carries through.

What is persona prompting?▼

Persona prompting assigns the AI a specific role or identity using "Act as a…" at the start. This shapes the AI's knowledge framing, vocabulary, and approach for the entire conversation. It's one of the simplest and highest-impact techniques — even beginners see immediate improvement.

Continue learning

Next steps

Apply what you've learned with real examples, or see the most common mistakes to avoid.

📚 Back to tutorial overview 📋 Explore real prompt examples 🚫 Common prompting mistakes

Advanced Prompting

Explore real prompt examples next