Write a How-To Guide AI Engines Will Cite
Generate a complete how-to / tutorial article structured exactly the way Google AI Overviews, ChatGPT Search, and Perplexity reward — answer-first intro, numbered atomic steps with image cues, prerequisites, time estimate, and a paste-ready HowTo JSON-LD block. Optimized for citation, not just ranking.
Pro tip
AI engines cite tutorials more often than any other content type because the question ("how do I…") maps cleanly to the answer (the steps). The trick is to write atomic steps — one action per step, not "Step 3: configure your account, choose a plan, and invite your team" — and to include a one-sentence answer to the implied question in the FIRST paragraph, before any "in this guide we'll cover…" filler.
How to use this prompt
- Pick your AI model. Choose the tab for Claude, ChatGPT, Gemini or Copilot — each variant is tuned for that model.
- Copy the full prompt. Click Copy Full Prompt to copy the text to your clipboard.
- Paste into your AI tool. Open your chosen model and paste the prompt into a new chat.
- Replace the
[placeholders]. Swap any bracketed fields for your company name, audience, product or tone. - Run and refine. Review the output. If anything is off, ask the AI to tighten tone, length or format.
Prompt Variants by Model
Claude
Claude 4.x
FRESH APR 2026
You are a tutorial writer who specializes in content that AI answer engines cite. Write a complete how-to guide structured exactly the way Google AI Overviews, ChatGPT Search, Perplexity, and Claude...You are a tutorial writer who specializes in content that AI answer engines cite. Write a complete how-to guide structured exactly the way Google AI Overviews, ChatGPT Search, Perplexity, and Claude reward — answer-first intro, atomic numbered steps with image cues, prerequisites, time estimate, and a paste-ready HowTo JSON-LD block.
<topic_brief>
What the reader will learn to do (one verb-led sentence): [e.g. "set up a 13-week cash flow forecast", "recover an abandoned cart sequence", "verify your domain in Google Search Console"]
Verbatim search query the reader is typing into ChatGPT or Google: [e.g. "how to set up cash flow forecast for small business"]
Reader''s skill level: [Beginner / Intermediate / Expert]
Reader''s context (what they have already / what they''re using): [e.g. "QuickBooks Online + a Google Sheet", "Shopify store with Klaviyo connected", "WordPress site with no plugins yet"]
Estimated total time to complete (e.g. "30 minutes"): [TIME or BLANK to let you estimate]
Estimated cost (USD, "0" if free): [AMOUNT]
Tools required (software, services): [LIST]
Supplies / assets required (files, accounts, data): [LIST]
Brand voice: [warm-expert / no-nonsense / playful / premium-restrained / technical-precise / friendly-everyday]
Author name + URL (for byline + schema): [NAME], [URL]
Specific proof point or credibility marker for the intro: [e.g. "we''ve done this for 200+ shops", "based on the official Stripe docs", "tested with 12 SMB clients in 2025"]
</topic_brief>
<structure_rules>
The guide MUST follow this exact structure:
1. **H1 title** — must contain the verb + the entity from the search query, ≤12 words.
2. **Answer-first intro paragraph** (≤80 words):
- First sentence directly answers the implied question in 20–40 words.
- Includes the proof point.
- No "in this guide we''ll cover…" filler.
- Ends by naming the time + skill level required so readers can self-qualify.
3. **Quick reference box** (just below intro): time required · skill level · cost · tools needed (bulleted list, ≤6 items).
4. **Prerequisites section** — atomic checklist of what must be true BEFORE step 1.
5. **Numbered steps** — 5–10 atomic steps. Each step:
- Imperative H2 heading starting with a verb (e.g. "## Step 3 — Connect your Stripe account")
- 1–3 sentence text. Self-contained: no "this", "it", "we" requiring prior context.
- One image cue in italics (e.g. *[Screenshot: Stripe API keys page with the test/live toggle highlighted]*).
- Optional inline code block if technical.
6. **Verify it worked** section — 2–4 sentence test the reader can run to confirm success.
7. **What can go wrong** section — 3–5 bullet troubleshooting items, each one-sentence symptom + one-sentence fix.
8. **What to do next** section — 1–3 bullets pointing to logical follow-up tutorials.
</structure_rules>
Think through the workflow step by step before writing — sequence the steps in execution order, identify where a non-expert would actually get stuck, and make sure no step assumes knowledge from an earlier one.
Output exactly four parts:
**PART 1 — The Guide**
The complete article in markdown, following the structure above. Image cues only (don''t invent URLs). Keep total length to 800–1500 words.
**PART 2 — HowTo JSON-LD (paste-ready)**
A complete <script type="application/ld+json"> block with @type HowTo: name, description (use the answer-first intro), totalTime (ISO 8601 duration), estimatedCost, supply (HowToSupply array), tool (HowToTool array), step (HowToStep array — each with name + text; image only if I provide one later), datePublished (today), author. Validate the JSON parses.
**PART 3 — Image Brief**
A list of every image cue in the guide, with: filename suggestion (kebab-case), alt text (descriptive, ≤120 chars), and what the image should show. Hand this to a designer or use it as a screenshot checklist.
**PART 4 — Pre-Publish Audit**
Quick checklist confirming:
- ✓/✗ H1 contains the verb + entity from the search query
- ✓/✗ Intro answers the implied question in ≤40 words of its first sentence
- ✓/✗ Every step is atomic (one action only)
- ✓/✗ No step uses "this", "it", or "we" without a clear antecedent
- ✓/✗ HowTo JSON-LD is valid
- Any field where you guessed at facts I should verify before publishing.
Notes: Claude is the strongest variant for the structural rules — it adheres to the atomic-step requirement most rigorously. Best for tutorials with 7+ steps.
ChatGPT
GPT-5
FRESH APR 2026
Act as a tutorial writer for content AI answer engines cite. Write a complete how-to guide structured the way Google AI Overviews, ChatGPT Search, and Perplexity reward — answer-first intro, atomic...Act as a tutorial writer for content AI answer engines cite. Write a complete how-to guide structured the way Google AI Overviews, ChatGPT Search, and Perplexity reward — answer-first intro, atomic numbered steps with image cues, prerequisites, time estimate, and a paste-ready HowTo JSON-LD block.
Topic brief:
- What the reader will learn to do (verb-led sentence): [e.g. "set up a 13-week cash flow forecast"]
- Verbatim search query: [e.g. "how to set up cash flow forecast for small business"]
- Reader skill level: [Beginner / Intermediate / Expert]
- Reader context (what they have / what they''re using): [DESCRIPTION]
- Estimated total time (e.g. "30 minutes"): [TIME or BLANK]
- Estimated cost (USD, "0" if free): [AMOUNT]
- Tools required (software, services): [LIST]
- Supplies / assets required: [LIST]
- Brand voice: [warm-expert / no-nonsense / playful / premium / technical / friendly-everyday]
- Author name + URL: [NAME], [URL]
- Specific proof point for the intro: [PROOF]
Required structure:
1. H1: verb + entity from query, ≤12 words.
2. Answer-first intro (≤80 words): first sentence answers the question in 20–40 words, includes the proof point, ends by naming time + skill level.
3. Quick reference box: time / skill / cost / tools (≤6 bullets).
4. Prerequisites: atomic checklist.
5. 5–10 numbered atomic steps. Each: imperative H2 starting with a verb; 1–3 sentence text; self-contained (no "this"/"it"/"we" without antecedent); one image cue in italics.
6. Verify it worked: 2–4 sentences.
7. What can go wrong: 3–5 troubleshooting bullets.
8. What to do next: 1–3 bullets.
Output:
1. The guide in markdown. 800–1500 words. Image cues only — don''t invent URLs.
2. Complete <script type="application/ld+json"> HowTo block: name, description (the intro), totalTime ISO duration, estimatedCost, supply, tool, step (name+text), datePublished today, author. JSON-valid.
3. Image brief: per cue — filename (kebab-case), alt text (≤120 chars), what the image should show.
4. Pre-publish audit checklist: ✓/✗ on each structural rule + flag guessed facts to verify.
Notes: GPT-5 sometimes pads steps with extra context — re-prompt with "split into atomic steps" if any step has more than one verb.
Gemini
Gemini 2.5 Pro
FRESH APR 2026
You are a tutorial writer for content AI answer engines cite. Write a complete how-to guide structured the way Google AI Overviews, Gemini, ChatGPT Search, and Perplexity reward — answer-first intro,...You are a tutorial writer for content AI answer engines cite. Write a complete how-to guide structured the way Google AI Overviews, Gemini, ChatGPT Search, and Perplexity reward — answer-first intro, atomic steps with image cues, prerequisites, time estimate, and paste-ready HowTo JSON-LD.
TOPIC BRIEF:
• What reader will learn (verb-led sentence): [DESCRIPTION]
• Verbatim search query: [QUERY]
• Reader skill level: [Beginner / Intermediate / Expert]
• Reader context: [DESCRIPTION]
• Total time: [TIME or BLANK]
• Cost (USD, 0 if free): [AMOUNT]
• Tools required: [LIST]
• Supplies required: [LIST]
• Brand voice: [warm-expert / no-nonsense / playful / premium / technical / friendly-everyday]
• Author + URL: [NAME], [URL]
• Specific proof point for intro: [PROOF]
REQUIRED STRUCTURE:
1. H1: verb + entity from query, ≤12 words.
2. Answer-first intro (≤80 words): sentence 1 answers question in 20–40 words, includes proof point, ends with time + skill required.
3. Quick reference box: time / skill / cost / tools (≤6 bullets).
4. Prerequisites: atomic checklist.
5. 5–10 numbered atomic steps. Each: imperative H2 verb-led; 1–3 sentence self-contained text; one italic image cue.
6. Verify it worked: 2–4 sentences.
7. What can go wrong: 3–5 troubleshooting bullets.
8. What to do next: 1–3 bullets.
OUTPUT:
1. Guide in markdown, 800–1500 words. Image cues only — no invented URLs.
2. Complete <script type="application/ld+json"> HowTo block: name, description, totalTime ISO duration, estimatedCost, supply, tool, step (name+text), datePublished today, author. JSON-valid.
3. Image brief: per cue — filename (kebab-case), alt text (≤120 chars), what to show.
4. Pre-publish audit: ✓/✗ on each rule + guessed-facts flag.
Notes: Gemini 2.5 Pro is excellent for the prerequisites + verify-it-worked sections. It tends to surface failure modes other models miss.
Copilot
Copilot (GPT-5)
FRESH APR 2026
Write a complete how-to / tutorial guide structured the way Google AI Overviews, ChatGPT Search, Perplexity, and Bing Copilot reward — answer-first intro, atomic numbered steps with image cues,...Write a complete how-to / tutorial guide structured the way Google AI Overviews, ChatGPT Search, Perplexity, and Bing Copilot reward — answer-first intro, atomic numbered steps with image cues, prerequisites, time estimate, and paste-ready HowTo JSON-LD.
Topic brief:
• What reader will learn (verb-led sentence): [DESCRIPTION]
• Verbatim search query: [QUERY]
• Reader skill level: [Beginner / Intermediate / Expert]
• Reader context: [DESCRIPTION]
• Total time: [TIME or BLANK]
• Cost (USD, 0 if free): [AMOUNT]
• Tools required: [LIST]
• Supplies required: [LIST]
• Brand voice: [warm-expert / no-nonsense / playful / premium / technical / friendly-everyday]
• Author + URL: [NAME], [URL]
• Specific proof point for intro: [PROOF]
Required structure:
1. H1: verb + entity from query, ≤12 words.
2. Answer-first intro (≤80 words): sentence 1 answers question in 20–40 words, includes proof point, ends with time + skill.
3. Quick reference box: time / skill / cost / tools (≤6 bullets).
4. Prerequisites: atomic checklist.
5. 5–10 numbered atomic steps. Each: imperative H2 verb-led; 1–3 sentence self-contained text; one italic image cue.
6. Verify it worked: 2–4 sentences.
7. What can go wrong: 3–5 troubleshooting bullets.
8. What to do next: 1–3 bullets.
Output:
1. Guide in markdown, 800–1500 words. Image cues only — no invented URLs.
2. Complete <script type="application/ld+json"> HowTo block: name, description, totalTime ISO duration, estimatedCost, supply, tool, step (name+text), datePublished today, author. JSON-valid.
3. Image brief: per cue — filename (kebab-case), alt text (≤120 chars), what the image should show.
4. Pre-publish audit checklist + guessed-facts flag.
Notes: Copilot is reliable on this prompt — particularly strong for tutorials targeting Microsoft-stack users (Power BI, M365, Azure).
Frequently Asked Questions
What does the Write a How-To Guide AI Engines Will Cite prompt do?
Generate a complete how-to / tutorial article structured exactly the way Google AI Overviews, ChatGPT Search, and Perplexity reward — answer-first intro, numbered atomic steps with image cues, prerequisites, time estimate, and a paste-ready HowTo JSON-LD block. Optimized for citation, not just ranking.
Which AI models is this prompt tested on?
This prompt is field-tested on Claude, ChatGPT, Gemini and Copilot. Each model has its own optimized variant above.
Do I need a paid AI account to use this prompt?
No. This prompt is written to run on the free tier of Claude, ChatGPT, Gemini and Copilot. Paid tiers simply give you longer context windows and faster responses.
Can I customize this prompt for my business?
Yes. Any text inside square brackets is a placeholder you replace with your own business details, such as company name, audience, product or tone. You can also ask the AI to adjust format, length or style after the first output.
When was this prompt last verified?
Each model variant above shows its own freshness stamp. AlignAI re-verifies every prompt at least monthly and rebuilds when a major model changes.
Don’t see what you need? tailored to your use case.