Turn Any How-To Blog Post into HowTo Schema
Paste a how-to or tutorial blog post and get back a complete, paste-ready Schema.org HowTo JSON-LD block — extracted steps, tools, supplies, total time, and image references — plus a content-quality audit of any step that AI engines won't cite confidently.
Pro tip
HowTo schema is the highest-leverage structured data on tutorial content for small businesses — it's actively consumed by Google AI Overviews and ChatGPT Search for "how do I..." queries. The catch: HowTo requires every step to have a name, a clear text, and ideally an image and url. Most blog posts have the steps; few have the structure AI engines need.
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 structured-data engineer. I''m going to paste a how-to or tutorial blog post. Your job: extract its steps, tools, supplies, time, and images into a complete Schema.org HowTo JSON-LD block —...You are a structured-data engineer. I''m going to paste a how-to or tutorial blog post. Your job: extract its steps, tools, supplies, time, and images into a complete Schema.org HowTo JSON-LD block — and audit any step that AI engines wouldn''t cite confidently.
<post_context>
Post URL: [https://YOUR-DOMAIN.COM/blog/...]
Author name: [AUTHOR]
Author URL (optional, e.g. /about): [URL or BLANK]
Date published (YYYY-MM-DD): [DATE]
Date last updated (YYYY-MM-DD): [DATE or SAME]
Featured image URL (the post''s hero image): [https://...]
Estimated total time to complete (e.g. "PT2H30M" or "30 minutes"): [TIME or BLANK — I''ll let you estimate]
Estimated cost (USD, optional, e.g. "USD 0" for free, "USD 50"): [AMOUNT or BLANK]
</post_context>
<post_content>
[PASTE THE FULL POST — headings, body, image references, captions. Markdown or HTML both fine.]
</post_content>
<extraction_workflow>
Walk through the post step by step:
1. Identify every distinct step in the tutorial. Number them in execution order.
2. For each step, extract:
- A short imperative name (e.g. "Mix the dry ingredients", "Install the dependency")
- A 1–3 sentence text (what to do)
- The image URL if one is present in that section (else null — DO NOT invent URLs)
- A heading anchor URL if the post uses anchors (else null)
3. Identify every tool mentioned (software, hardware, external services).
4. Identify every supply mentioned (materials, ingredients, datasets, files).
5. Estimate total time if I didn''t supply one — base it on step count and typical complexity.
6. Identify the implicit question the tutorial answers and craft a clean `name` + `description`.
</extraction_workflow>
Output exactly four sections:
**PART 1 — Extraction Summary**
- Number of steps detected
- Total tools, total supplies
- Estimated total time (with reasoning if estimated)
- Featured image confirmed (yes/no/missing)
- Implicit question this tutorial answers
**PART 2 — Step-by-Step Quality Audit**
Markdown table: Step # · Name · Has clear text? (y/n) · Has image? (y/n) · Has anchor URL? (y/n) · AEO-ready? (y/n + one-line issue if no).
For any step marked NOT ready, suggest a one-line fix (e.g. "split into 2 steps", "needs an image — recommend a screenshot of the result", "rewrite — currently uses ''this'' without antecedent").
**PART 3 — HowTo JSON-LD (paste-ready)**
A complete <script type="application/ld+json"> block with @type HowTo, including: name, description, image, totalTime (ISO 8601 duration), estimatedCost, supply (HowToSupply array), tool (HowToTool array), step (HowToStep array — each with name, text, image if present, url if present), datePublished, dateModified, author. Validate the JSON parses. Use null only where it''s structurally valid; otherwise OMIT the field.
**PART 4 — Deploy & Verify**
- Where to paste in the post''s <head>
- Rich Results Test URL: https://search.google.com/test/rich-results?url=[POST-URL]
- A note: if I keep visible numbered steps in the body copy, do NOT add HowTo microdata to those steps too — JSON-LD only, never both.
- Flag any field where you had to guess or estimate that I should verify before publishing.
Notes: Claude handles long blog posts well and produces the cleanest step extraction. Best variant for posts with 8+ steps.
ChatGPT
GPT-5
FRESH APR 2026
Act as a structured-data engineer. I''ll paste a how-to / tutorial blog post. Extract steps, tools, supplies, time, images into a complete Schema.org HowTo JSON-LD block. Audit any step AI engines...Act as a structured-data engineer. I''ll paste a how-to / tutorial blog post. Extract steps, tools, supplies, time, images into a complete Schema.org HowTo JSON-LD block. Audit any step AI engines won''t cite confidently.
Post context:
- URL: [https://YOUR-DOMAIN.COM/blog/...]
- Author: [NAME]
- Author URL (optional): [URL or BLANK]
- Date published (YYYY-MM-DD): [DATE]
- Date updated (YYYY-MM-DD): [DATE or SAME]
- Featured image URL: [https://...]
- Total time estimate (ISO duration "PT2H30M" or plain): [BLANK to let you estimate]
- Cost (USD, optional): [AMOUNT or BLANK]
Post content (paste full — markdown or HTML):
[PASTE]
Workflow:
1. Identify every distinct step. Number in execution order.
2. For each: short imperative name, 1–3 sentence text, image URL if present (else null — don''t invent), anchor URL if present.
3. Identify all tools (software, hardware, services).
4. Identify all supplies (materials, ingredients, datasets, files).
5. Estimate total time if missing.
6. Identify the implicit question and craft `name` + `description`.
Output:
1. Extraction Summary: step count, tool count, supply count, estimated time + reasoning, featured image confirmed (y/n), implicit question.
2. Step-by-step audit table: # · Name · Has text (y/n) · Has image (y/n) · Has anchor (y/n) · AEO-ready (y/n + one-line issue). Add a one-line fix for any NOT ready.
3. Full <script type="application/ld+json"> HowTo block: name, description, image, totalTime, estimatedCost, supply, tool, step (each with name/text/image/url), datePublished, dateModified, author. JSON-valid. Omit fields rather than null where valid.
4. Deploy: paste location, Rich Results Test URL, microdata-vs-JSON-LD warning, flag any guessed/estimated values to verify.
Notes: GPT-5 occasionally invents image URLs from filenames in the markdown — verify every image URL it returns actually exists.
Gemini
Gemini 2.5 Pro
FRESH APR 2026
You are a structured-data engineer. I''ll paste a how-to / tutorial blog post. Extract its steps, tools, supplies, time, and images into a complete Schema.org HowTo JSON-LD block. Audit any step AI...You are a structured-data engineer. I''ll paste a how-to / tutorial blog post. Extract its steps, tools, supplies, time, and images into a complete Schema.org HowTo JSON-LD block. Audit any step AI engines won''t cite confidently.
POST CONTEXT:
• URL: [https://YOUR-DOMAIN.COM/blog/...]
• Author: [NAME]
• Author URL: [URL or BLANK]
• Date published (YYYY-MM-DD): [DATE]
• Date updated: [DATE or SAME]
• Featured image URL: [https://...]
• Total time (ISO "PT2H30M" or plain): [BLANK to estimate]
• Cost (USD): [AMOUNT or BLANK]
POST CONTENT (paste full markdown or HTML):
[PASTE]
WORKFLOW:
1. Identify every step in execution order.
2. For each: imperative name, 1–3 sentence text, image URL if present (else null — don''t invent), anchor URL if present.
3. Identify tools (software, hardware, services).
4. Identify supplies (materials, ingredients, datasets, files).
5. Estimate total time if missing.
6. Identify implicit question; craft `name` + `description`.
OUTPUT:
1. Extraction Summary: step count, tool/supply counts, estimated time + reasoning, featured image (y/n), implicit question.
2. Step audit table: # | Name | Has text | Has image | Has anchor | AEO-ready (y/n + one-line issue). Add a one-line fix for each NOT ready.
3. Complete <script type="application/ld+json"> HowTo block — JSON-valid, name, description, image, totalTime, estimatedCost, supply, tool, step (name/text/image/url), datePublished, dateModified, author. Omit fields instead of null where valid.
4. Deploy: paste location in <head>, Rich Results Test URL, microdata-vs-JSON-LD warning, flag any guessed values.
Notes: Gemini 2.5 Pro is strong at the step-quality audit. Particularly good at flagging steps that need image references.
Copilot
Copilot (GPT-5)
FRESH APR 2026
Extract a how-to / tutorial blog post into a complete Schema.org HowTo JSON-LD block, plus a step-quality audit.
Post context:
• URL: [https://YOUR-DOMAIN.COM/blog/...]
• Author: [NAME]
• Author...Extract a how-to / tutorial blog post into a complete Schema.org HowTo JSON-LD block, plus a step-quality audit.
Post context:
• URL: [https://YOUR-DOMAIN.COM/blog/...]
• Author: [NAME]
• Author URL: [URL or BLANK]
• Date published (YYYY-MM-DD): [DATE]
• Date updated: [DATE or SAME]
• Featured image URL: [https://...]
• Total time (ISO "PT2H30M" or plain): [BLANK to estimate]
• Cost (USD): [AMOUNT or BLANK]
Post content (paste full markdown or HTML):
[PASTE]
Workflow:
1. Identify every step in execution order.
2. For each step: imperative name, 1–3 sentence text, image URL if present (else null — don''t invent), anchor URL if present.
3. Identify tools (software, hardware, services).
4. Identify supplies (materials, ingredients, datasets, files).
5. Estimate total time if missing.
6. Identify implicit question; craft `name` + `description`.
Output:
1. Extraction Summary: step count, tool/supply counts, estimated time + reasoning, featured image (y/n), implicit question.
2. Step audit table: # · Name · Has text (y/n) · Has image (y/n) · Has anchor (y/n) · AEO-ready (y/n + one-line issue). Add a one-line fix for each NOT ready.
3. Full <script type="application/ld+json"> HowTo block — JSON-valid; name, description, image, totalTime, estimatedCost, supply, tool, step (name/text/image/url), datePublished, dateModified, author. Omit fields instead of null where valid.
4. Deploy: paste location, Rich Results Test URL, microdata-vs-JSON-LD warning, flag any guessed values.
Notes: Copilot is solid for this prompt. Strongest fit when the post lives on a Microsoft-stack site (SharePoint, Word-published) where Copilot can also render the resulting Rich Result preview.
Frequently Asked Questions
What does the Turn Any How-To Blog Post into HowTo Schema prompt do?
Paste a how-to or tutorial blog post and get back a complete, paste-ready Schema.org HowTo JSON-LD block — extracted steps, tools, supplies, total time, and image references — plus a content-quality audit of any step that AI engines won't cite confidently.
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.