AI AI Tools Guidebook
Prompt Library

Blog Outline Prompts: 17 Templates for Articles That Rank in 2026

17 copy-ready blog outline prompts for SEO posts, pillar pages, comparisons, and cluster content — built for Google's 2026 AI Overviews, where self-contained passages and question headings win the citation.

Published: Updated: Author: AI Productivity Guide Team 🌐 查看中文版本

Blogs fail at the outline. If you skip it and ask AI for a 1,500-word article, you get bland filler that Google’s March 2026 core update treats as exactly that. The fix is to make the model write the outline first, then attack each section with its own prompt. A good outline is also an SEO spec: as of June 2026, Google AI Overviews extract self-contained passages of roughly 134–167 words built under question-style headings, and about 44% of AI-search citations come from the first 30% of a page. An outline that front-loads the answer and slices the article into standalone blocks is what gets cited.

Outlining a long-form explainer on a historical topic? Pair these with the AI history-timeline workflow so the outline is anchored by real dates, actors, and causal chains rather than generic H2 patterns.

TL;DR

  • Outline first, draft second. Spend 200 words on structure, not 1,500 on filler you’ll cut.
  • Every prompt below must name six things: target query, audience, article type, format constraints, a per-section takeaway, and a closing action.
  • For 2026 search, use question-format H3s and write each section as a self-contained ~150-word answer block — that is the unit AI Overviews lift.
  • Model pick (June 2026): GPT-5.5 for versatile, creative, and marketing posts; Claude Opus 4.7 for accuracy-critical explainers and dense source material. Both carry a 1M-token context, so you can paste competitor SERPs and source docs straight in.
  • Reserve one section for what no competitor covers — Google’s 2026 updates reward information originality, not SERP mirroring.

What these prompts solve

A blog outline prompt has one job: turn an intent (rank for X / teach Y / convince Z) into a section-by-section skeleton with explicit search-intent slots, evidence slots, and a takeaway. Once the outline is right, drafting is mechanical, and you can swap models or human writers without losing structure. (Outlining a 45-minute podcast episode instead of an article? Use the AI podcast outline workflow — hook, five segments with time markers, transitions, audience questions — built for the listening medium rather than scanning.)

This matters more in 2026 than it did two years ago. Google’s March 2026 core update re-weighted three signals: information originality (does the page say anything that exists nowhere else), demonstrable author expertise, and topical coherence over time. AI assistance is fine — Google does not penalize it categorically — but a generic AI skeleton with no original angle now ranks like the filler it is. The outline is where you bake in the angle.

Which model to use for outlining (June 2026)

For most outlining and drafting work, GPT-5.5 (ChatGPT’s default since late April 2026) is the strongest broad generalist — fast at brainstorming structure, alternate angles, and turning notes into draftable material. Reach for Claude Opus 4.7 when the article is accuracy-critical (technical explainers, finance, legal, anything where a wrong claim costs you), because it is more cautious about unverified claims. Both ship a 1M-token context window, so the SERP-paste and source-doc templates below work without truncation. On the free tiers you can outline fine; Plus ($20/mo) or Claude Pro ($20/mo) mainly buy you higher limits and longer thinking, not a different outline quality.

Who this is for, and when to skip these prompts

Content marketers building pillar pages and cluster content, indie devs writing SEO posts for their own products, founders shipping launch copy that has to convince and convert, and technical writers turning internal docs into public articles — anyone who keeps drafting 1,200-word posts that read like AI slop.

Skip the outline overhead for short news updates and changelog entries; the structure costs more than the post. Skip it for thought-leadership pieces where the structure is the argument (a contrarian essay’s outline is a thesis, not a list of H2s). And don’t outline before you have the search-intent keyword — without a target query, the model produces a generic article every time. If you don’t even have a topic yet, run the 30-angles-in-10-minutes brainstorm workflow first to surface candidates worth outlining.

Prompt anatomy: the six things every outline prompt must name

A blog outline prompt should always specify six things. Drop any one and the model defaults to a generic shape:

  • Target query / intent: the keyword or question the article answers.
  • Audience: who reads it, and their existing knowledge level.
  • Article type: tutorial, comparison, pillar, listicle, opinion, or problem-solution.
  • Format constraints: H2 count, total word budget, must-have sections (FAQ, comparison table, decision tree).
  • Insight slot per section: each H2 ships one specific takeaway, not generic prose.
  • Closing action: what the reader should do or decide by the end.

Add one more constraint for 2026 search: ask the model to make each section a self-contained answer block of roughly 150 words under a question-format subhead. That is the exact unit Google AI Overviews and ChatGPT search lift when they cite a source, and it forces the outline to front-load real answers instead of warming up for three paragraphs.

These templates cover the article types that benefit most: SEO tutorials, “X vs Y” comparisons, pillar pages anchoring a cluster, problem-solution posts, listicles that earn the list, framed personal-experience essays, and newsletter versions of long posts.

17 copy-ready prompt templates

1. Search-intent outline (keyword in hand)

You are a senior content editor. Write a blog outline for the keyword "{keyword}". Audience: {audience}. Format: H1, 5–7 H2 sections, 2–3 H3 sub-questions per H2. For each section: one sentence on the specific insight it delivers. End with a 3-bullet "what the reader can do now" block.

2. Pillar-page outline

Draft a pillar article outline on "{topic}". Goal: become the definitive resource. Include: (a) one-sentence definition, (b) 7–10 sub-sections each addressing one common question, (c) a comparison table section, (d) FAQ section with 5 questions, (e) related-topics list at the end with hub-style internal-link slots.

3. Compare-and-decide outline

I am writing an article comparing {A} vs {B}. Produce an outline that helps the reader decide: H1, "TL;DR — pick X if…" intro, 4 dimension sections (e.g., price, ease, features, support), real-world scenario examples, and a final decision matrix.

4. Listicle that earns the list

Outline a listicle "Top {N} {thing} for {audience}." Each item must have: (a) one-line summary, (b) best use case, (c) one downside, (d) who should skip it. Open with a 3-paragraph intro explaining how you ranked. Close with a "how to choose" decision tree.

5. Beginner tutorial outline

Outline a beginner tutorial on "{task}". Assume the reader has never done this before. Include: prerequisites, glossary of 5 must-know terms, numbered steps (5–9), what success looks like, 4 common mistakes, troubleshooting FAQ.

6. “X vs Y” SERP-fit outline

Write an outline for "{A} vs {B}: Which Should You Choose in 2026?". Structure: 2-sentence verdict up top, side-by-side comparison table (5 dimensions), 4 sections covering use cases where each wins, migration notes, FAQ. Mirror the H2 patterns of the top 3 SERP results — I'll paste them below.

SERP H2s: {paste}

7. How-to outline with skimmable subheads

Outline a how-to article on "{task}". Subheads should be question-format (e.g., "What do I need before I start?"). Each subhead followed by 2–4 sentences and one concrete example. Total 8–10 subheads.

8. Opinion / takedown outline

I want to write a strong-opinion article: "{thesis}". Outline it: hook with a counter-intuitive claim, 3 supporting arguments with evidence slots, 1 strongest objection + response, 1 honest concession, decisive closing. End with one-sentence reframe of the title.

9. Personal-story outline

Outline a personal-story blog about {experience}. Structure: hook scene, the inciting decision, what I tried, what failed, what worked, the principle I extracted, what I'd do differently. Each section 100–200 words.

10. Newsletter-from-blog outline

I have a 1,500-word blog post draft (pasted below). Outline a 400-word newsletter version: one hook line, 3 takeaways, one personal note, one CTA. Cut all marketing fluff.

{paste blog}

11. Cluster article outline (links up to a pillar)

Outline a cluster article on "{narrow-topic}" that supports the pillar "{pillar-topic}". The article must: (a) answer one specific question the pillar only glances at, (b) reference 2 sibling cluster articles by topic, (c) include one explicit upward link to the pillar in the intro and one in the conclusion. Output the outline with internal-link anchors marked [[link to: pillar]] and [[link to: sibling X]].

12. Problem-solution outline

Outline a problem-solution article for the reader query "{paste exact query}". Structure: 1-paragraph framing of the pain, 3 wrong answers (why they fail), the working answer (3 H2s of actual implementation), the verification step, the next problem this unlocks.

13. Update-and-refresh outline

I have an article from {year} on "{topic}" (paste below). Outline the refreshed version for 2026: which sections survive as-is, which need a data refresh, which need to be rewritten, which to drop entirely, what to add that wasn't relevant before. Show the new section order.

Old article: {paste}

14. SERP-gap outline

I want to rank for "{keyword}". Below are the top 5 results' H2s. Outline an article that covers everything they cover PLUS one section addressing a gap none of them fill. Identify the gap explicitly.

Top 5 H2s: {paste}

15. Tutorial outline with code-block plan

Outline a code tutorial on "{task}" in {language}. For each step: H2 title (≤8 words), what the reader does, the exact line(s) of code (just a placeholder — `// step N code here`), and the expected output. End with a "common errors and fixes" section listing at least 5 entries.

16. Outline for an under-defined topic

The topic "{topic}" is fuzzy — readers search it for very different reasons. Identify the 3 dominant search intents. Pick ONE and outline an article that owns that intent (don't try to serve all three). State which intent you picked and which two you deliberately excluded.

17. Anti-listicle outline

Outline an article on "{topic}" that deliberately is NOT a listicle. Argument-driven prose, 4–5 H2s, each carrying one claim with evidence. Length: 1,200 words. Title must NOT contain a number. Goal: rank without joining the "Top N" SERP slop.

Common mistakes

  • Asking for the full article before nailing the outline. You spend a token budget on 1,500 words of filler instead of 200 words of structure.
  • No target keyword or audience in the prompt. The model produces a generic shape that fits every topic and ranks for none.
  • Skipping the “decision / takeaway” section. Readers leave without knowing what to do; bounce rate climbs.
  • One outline for every article type. Tutorials, comparisons, and listicles need different shapes; reusing one template flattens them.
  • Outline that mirrors competitors exactly. You end up #6 on a SERP of identical articles. Always reserve one section for what no one else covers.
  • Not using H3 questions. Skimmable subheads are how readers (and Google) find their answer; H2-only outlines hide the structure.

How to push results further

  • Paste the top 3–5 SERP H2 patterns into the prompt. With a 1M-token context on GPT-5.5 and Opus 4.7, you can drop in the full competing pages — even a one-line summary per result helps the model match intent without copying it.
  • Add a takeaway line per H2 inside the outline. If a section can’t ship one specific takeaway, cut it before you draft.
  • Use H3 questions instead of statements. They map directly to People Also Ask and to the question-headings that AI Overviews favor.
  • For pillar pages, mark internal-link slots in the outline ([[link to: cluster X]]) so the model plans the cluster, not just the article.
  • For comparisons, fix the dimension count up front (4 or 5) and demand one verdict line per dimension. That forces ranking instead of “both have pros and cons” mush.
  • After the outline, ask the model to list one fact to verify per section before drafting. This surfaces hallucination risk before you spend tokens on the draft — non-negotiable now that Google’s 2026 updates weight factual originality.
  • If an outline reads generic, regenerate with one extra constraint per turn (audience seniority, format, dominant intent). Generic output comes from missing constraints, not a weak model.

FAQ

  • Which model should I outline with in 2026? GPT-5.5 for versatile, creative, and marketing posts; Claude Opus 4.7 for accuracy-critical explainers and dense source documents. Free tiers handle outlining fine; Plus or Claude Pro (both $20/mo as of June 2026) mainly raise your usage limits.
  • Should I write the outline in markdown or in the prompt itself? Markdown. It composes with the next prompt (drafting), and you can paste sections one at a time. The outline is the spec; the draft is the artifact.
  • Why does my model keep suggesting the same 7 H2s for every topic? You didn’t pin search intent. Try the SERP-gap template (#14) — forcing the model to look at gaps breaks the generic pattern.
  • How do I structure an outline so AI Overviews actually cite it? Make every section a self-contained ~150-word answer under a question-format subhead, and put the direct answer in the first sentence. AI search extracts short standalone passages and front-loads the top of the page, so a warm-up intro before the answer rarely gets pulled. Google’s own creating helpful content guidance is the source of truth for what “people-first” means.
  • How long should the outline be? ~200–400 words for a 1,500-word article. If your outline is longer than your draft will be, you over-outlined; cut to top-level structure.
  • Can I outline a 5,000-word pillar in one prompt? Yes for the skeleton, but draft section by section. One prompt = one section keeps quality high and lets you swap models per section.
  • How do I keep voice consistent across a multi-prompt drafting flow? Pin a 100-word voice spec at the top of every section prompt: tone, sentence-length target, banned phrases. Without it, voice flattens by section 3.

Tags: #Prompt #Writing #Blog outline #SEO

Related Articles