thought-leadership-writer · diff
git:20260504.0d54b6b to git:20260808.7dfb5c7
147 added, 14 removed. Audit A to A.
---
name: thought-leadership-writer
- description: Write long-form thought leadership articles, opinion pieces, industry POV essays, and CEO/founder bylines using the Made to Stick SUCCESs framework (Chip and Dan Heath). Use when the user asks for a long-form article, executive byline, opinion piece, industry POV, manifesto, "explain our point of view on X", or wants to publish an authority-building piece (1200-2500 words). Reads brand voice and positioning from knowledge/.
- reads:
- - knowledge/brand/voice.md
- - knowledge/markets/positioning.md
- - knowledge/markets/competitors.md
- - knowledge/icp/personas.md
- - knowledge/content-library/
- writes:
- - output/thought-leadership/
+ description: Write long-form thought leadership articles, opinion pieces, industry POV essays, and CEO/founder bylines using the Made to Stick SUCCESs framework (Chip and Dan Heath). Use when the user asks for a long-form article, executive byline, opinion piece, industry POV, manifesto, "explain our point of view on X", or wants to publish an authority-building piece (1200-2500 words). Reads brand voice and positioning from knowledge/. For a short-form post, see linkedin-post. For the newsletter, see newsletter-writer. For improving an existing draft, see copy-review.
+ metadata:
+ grounded_in:
+ - "Made to Stick - Heath & Heath"
+ reads:
+ - knowledge/brand/voice.md
+ - knowledge/markets/positioning.md
+ - knowledge/markets/competitors.md
+ - knowledge/icp/personas.md
+ - knowledge/content-library/
+ writes:
+ - output/thought-leadership/
---
# thought-leadership-writer
Writes long-form opinion pieces that establish authority and get shared. Uses the **Made to Stick SUCCESs framework** (Chip & Dan Heath, *Made to Stick*) to ensure every piece is memorable, not just well-written.
**The core insight from Made to Stick:** Ideas fail to stick not because they're wrong, but because they're abstract, predictable, and forgettable. The SUCCESs framework systematically fixes this. A thought leadership piece that checks all 6 conditions will be remembered, quoted, and shared. One that checks none will be read once and forgotten.
## The SUCCESs Framework
Apply all 6 conditions to every piece. Each one is a checklist item, not optional.
| Letter | Condition | The test | How to apply |
|---|---|---|---|
| **S** | Simple | Can you express the core idea in one sentence? | Find the lead - the single most important insight. Strip everything that doesn't serve it. |
| **U** | Unexpected | Does the piece break a pattern the reader expected? | Open by violating an assumption. Say the thing the reader didn't see coming. |
| **C** | Concrete | Are there specific people, places, objects, and numbers? | Replace abstract claims with tangible examples. "Activation rate stuck at 52%" not "low engagement". |
| **C** | Credible | Why should anyone believe this? | Use statistics, named examples, or an "anti-authority" (a customer or peer, not just the brand). |
| **E** | Emotional | Does the reader care? | Tap the "self-interest" trigger (this affects you) or the "identity" trigger (this is about who you are). |
| **S** | Stories | Is there a narrative? | Use a story - even a short one - to make the idea travel. People remember stories, not arguments. |
+ ### The villain the framework exists to defeat: the curse of knowledge
+
+ *Made to Stick* names one root cause for unstuck ideas. Once you know something, you cannot
+ imagine not knowing it, so you write the summary of your insight rather than the thing that
+ produced it. Every element of SUCCESs is a countermeasure to that one failure.
+
+ Practical test before you draft: hand the thesis sentence to someone outside the company. If they
+ need a follow-up question to understand what it claims, the curse is active and the piece will
+ read as abstract to everyone but the author.
+
+ ### Heath's credibility toolkit, when you have no statistic
+
+ Credible is the element most often skipped in B2B writing, usually because the author has no
+ study to cite. The Heaths list several sources of credibility that are not statistics. Pick one
+ deliberately rather than defaulting to a number you would have to invent.
+
+ | Technique | What it is | Use it when |
+ |---|---|---|
+ | Anti-authority | A source credible precisely because they have no incentive to say it: a customer, a skeptic, a competitor conceding a point | The brand claiming it would read as self-serving |
+ | Testable credential | An invitation for the reader to check for themselves: "look at your own funnel between step 6 and step 11" | The reader has the data on hand |
+ | Vivid detail | Specific, checkable texture that signals first-hand knowledge | The claim rests on the author having actually been there |
+ | Human-scale statistic | A number rescaled to something a person can picture | You have a real figure but it is too large to feel |
+ | Sinatra test | One example so strong it settles the question by itself | You have one undeniable case rather than many weak ones |
+
+ **Rule: a human-scale statistic still needs a real underlying number.** Rescaling is a
+ presentation move, not a licence to estimate one. If there is no number, use anti-authority or a
+ testable credential instead.
+
+ ### The three plots, for choosing the story rather than groping for one
+
+ | Plot | Shape | Best for a piece that argues |
+ |---|---|---|
+ | Challenge | Someone faces an obstacle bigger than them and overcomes it | This is hard but doable, here is the proof |
+ | Connection | Someone bridges a gap between people or groups | The problem is really a misalignment between two functions |
+ | Creativity | Someone solves a problem in a way nobody expected | The conventional method is not the only method |
+
+ Pick the plot from the argument, not from whichever anecdote is nearest. A Challenge plot
+ attached to a Creativity argument makes the piece feel effortful rather than surprising.
+
## When to use
- "Write a thought leadership piece on X"
- "Draft an opinion article about Y"
- "We need a POV piece on the future of Z"
- "Write a CEO byline for <publication>"
- "Write a manifesto"
## Inputs needed
- **Thesis** (required): the one-sentence claim this piece defends. If the user doesn't have one, help them sharpen one before writing.
- **Audience**: industry insiders, prospects, peers, press (default: prospects)
- **Length**: 1200, 1800, or 2500 words (default: 1800)
- **Author voice**: founder, CEO, CMO, or company POV
- **Publication target**: own blog, LinkedIn article, Substack, trade pub
## Process
### Step 1: Load context
- Read `knowledge/brand/voice.md`, `knowledge/markets/positioning.md`, `knowledge/markets/competitors.md`. Stop if voice is missing: "Run `/onboard --brand` first."
+ Read `knowledge/brand/voice.md`, `knowledge/markets/positioning.md`, `knowledge/markets/competitors.md`. Stop if voice is missing: "Run `/brand-context` first."
### Step 2: Sharpen the thesis
A weak thesis kills everything downstream. Before writing, run it through three tests:
**Test 1 - Specificity.** Can it be argued with?
- ✗ "AI will change marketing" (too vague)
- ✗ "Content marketing is important" (too obvious)
- ✓ "Activation rate is a proxy metric. Here's what actually predicts retention."
- ✓ "Most onboarding flows are too long. Cutting steps 6-11 will outperform redesigning steps 1-5."
**Test 2 - Contestability.** Would a reasonable person disagree?
- If everyone would agree, it is not thought leadership. It is a press release.
- The strongest pieces make people uncomfortable before they convince them.
**Test 3 - Alignment.** Does it reinforce `knowledge/markets/positioning.md`?
- Thought leadership should move the brand's position forward, not sideways.
- If the thesis contradicts the company's POV, flag it before writing.
+ **Score the thesis before proceeding.** One point each:
+
+ | Point | Condition |
+ |---|---|
+ | 1 | A competent person in this industry could disagree with it in one sentence |
+ | 2 | It names a specific mechanism, metric, step or population, not a trend |
+ | 3 | The author has first-hand material that supports it, not just an opinion |
+ | 4 | It moves `knowledge/markets/positioning.md` forward rather than sideways |
+
+ **Threshold: write at 3 or 4. At 2, propose 2-3 sharper versions and rewrite. At 0 or 1, stop and
+ say the piece should not be written yet**, because a thesis nobody can disagree with produces an
+ article nobody can remember, and no amount of SUCCESs work at later steps rescues it.
+
If the thesis fails any test, propose 2-3 sharper versions before proceeding.
### Step 3: Run the SUCCESs pre-write
Complete this before writing:
```
SUCCESs pre-write for: [Working title]
S - SIMPLE (the core)
One-sentence core idea (not the title - the insight):
What gets cut if we had to strip 50% of the piece:
U - UNEXPECTED (the pattern break)
What does the reader expect to hear on this topic?
What's the unexpected angle or opening move?
First sentence candidate (the pattern-breaker):
C - CONCRETE (the specifics)
3 specific examples, numbers, or named cases to use:
1.
2.
3.
Any abstract claims that need a concrete replacement?
C - CREDIBLE (the proof)
- Primary credibility source (stat, study, customer, or self-experience):
+ Credibility technique chosen (anti-authority | testable credential | vivid detail |
+ human-scale statistic | Sinatra test):
+ Primary credibility source (stat, study, customer, or self-experience) and where it came from:
Anti-authority candidate (a peer, customer, or skeptic who validates the point):
What the reader might object to, and the counter:
E - EMOTIONAL (why they should care)
Self-interest angle: how does this affect the reader directly?
Identity angle: what does believing this say about who they are?
The "curse of knowledge" check: are we assuming context the reader doesn't have?
S - STORIES (the narrative)
+ Plot type chosen (challenge | connection | creativity), and why it matches the argument:
Opening story candidate (specific scene, not a parable):
Supporting anecdote (from knowledge/content-library/ or user input):
The moment of realization (the turn in the narrative):
```
### Step 4: Structure the piece
Use this structure. Each section maps to SUCCESs elements.
```
## 1. Hook (150-200 words) [U + S + stories]
Open with the UNEXPECTED move. Options:
- A specific scene ("Last Tuesday, a Head of Growth showed me her dashboard...")
- A counterintuitive stat ("The average B2B trial has 11 onboarding steps. The best ones have 4.")
- A direct challenge to conventional wisdom ("Most companies are measuring the wrong metric.")
Rule: The thesis should appear by the end of the hook, or at the start of section 2.
Never open with "In today's..." / "As we all know..." / "It's no secret that..."
## 2. What everyone believes (200-300 words) [Credible]
Steelman the conventional wisdom. Be generous. Don't strawman.
- "The standard advice is X, and it's not wrong. It's just incomplete."
- Name real examples of the conventional approach being applied
## 3. Why it's wrong (or incomplete) (300-500 words) [Unexpected + Concrete + Credible]
The UNEXPECTED argument. Specific evidence.
- Name the cracks in the conventional approach
- Use concrete examples (named companies, specific numbers, real situations)
- "We've seen this in [X] clients. The pattern is..."
- The anti-authority: quote a peer, customer, or skeptic - not just the brand
## 4. The new model (400-600 words) [Simple + Concrete + Emotional]
The author's POV. This is where the SIMPLE core lives.
- Define new terms if needed (give the idea a name it can travel with)
- Show how the new model works with a CONCRETE example
- Tap EMOTIONAL: "Here's why this matters for you specifically"
- Include a framework, visual, or 3-part structure the reader can remember and repeat
## 5. What to do about it (200-400 words) [Concrete + Emotional + Stories]
Practical application. Not abstract implications.
- 3 specific actions the reader can take
- At least one they can take today or this week
- One supporting story from knowledge/content-library/ if available
## 6. Close (100-150 words) [Simple + Unexpected]
The quotable line. One sentence the reader will screenshot.
- Restate the core insight in its sharpest form
- Call to something: a belief, an action, a question
- The ending should feel inevitable in hindsight, not obvious from the start
```
### Step 5: Write
Apply voice from `knowledge/brand/voice.md`. Rules:
- Specific examples, named companies, real numbers throughout
- Cite any statistics with the source (inline link)
- One idea per paragraph - maximum 3 sentences
- Headers in sentence case
- No em dashes
- No: "In today's rapidly evolving landscape", "It's no secret that", "At the end of the day"
### Step 6: SUCCESs self-check (run after writing)
```
S - Simple: [ ] Core idea expressible in one sentence
[ ] No paragraphs that don't serve the core idea
U - Unexpected: [ ] Opening breaks a pattern or assumption
- [ ] Reader encounters at least 1 "I didn't expect that" moment
+ [ ] Quote the sentence that contradicts what section 2 said everyone believes.
+ If no sentence does, the piece agrees with the conventional wisdom it set up
- C - Concrete: [ ] At least 3 specific examples (named company, number, or scene)
+ C - Concrete: [ ] At least 3 specific examples, each traceable to `knowledge/` or user input. Any example you could not source is marked `[SOURCE NEEDED]` in the draft, never written as fact
[ ] No abstract claim left without a concrete anchor
C - Credible: [ ] Primary credibility source present (stat, case study, or experience)
[ ] At least 1 "anti-authority" (peer/customer validates, not just the brand)
E - Emotional: [ ] Piece taps self-interest ("this affects you") or identity ("this is about who you are")
[ ] No "curse of knowledge" - doesn't assume context reader doesn't have
S - Stories: [ ] At least 1 narrative (specific scene with a beginning, middle, and turn)
[ ] Close has a line worth quoting or sharing
+ Sourcing: [ ] Every statistic, quote, company name and result is listed with its source.
+ Anything without one is `[SOURCE NEEDED: ...]` or `[NEEDS INPUT: ...]`
+ [ ] Every external link was fetched in this session and resolved
+ [ ] The credibility technique named in the pre-write is the one actually used
+
Voice checks:
[ ] Thesis appears before word 300
[ ] Voice matches knowledge/brand/voice.md
- [ ] No em dashes, no filler openers, no buzzwords
+ [ ] Zero em dashes and zero en dashes
+ [ ] Zero occurrences of: "In today's rapidly evolving landscape", "It's no secret
+ that", "At the end of the day", leverage, unlock, robust, seamless
+ [ ] Word count within 10% of the target
```
### Step 7: Save
`output/thought-leadership/<DD-MM-YYYY>-<slug>.md` with frontmatter:
```yaml
---
format: thought-leadership
framework: made-to-stick-success
thesis: <one-sentence claim>
author: <name or role>
target: <publication>
words: <count>
success-score: S/U/C/C/E/S checked
created: DD-MM-YYYY
---
```
### Step 8: Offer derivative assets
- LinkedIn post version (`/linkedin-post`)
- Substack note (3-5 sentences, the sharpest version of the thesis)
- 5-tweet thread
- 3 pull-quote candidates for graphics (the most quotable lines)
## Rules
- Run the SUCCESs pre-write before writing. Pieces written without it will be technically correct and completely forgettable.
- A thought leadership piece must take a position someone reasonable could disagree with. If your draft offends no one, it is not thought leadership.
- Never invent quotes, stats, or examples. Flag "user should add a real source here" rather than fabricating.
- Named executive pieces: "Confirm this reflects what <Name> actually believes before publishing."
- The close must contain 1 quotable line. If you can't find it, rewrite the close.
## SUCCESs quick reference
| Most common failures | Fix |
|---|---|
| Piece is technically accurate but boring | Missing U (unexpected) and S (stories) |
| Piece makes big claims nobody believes | Missing C (credible) |
| Piece is smart but nobody shares it | Missing E (emotional) - reader doesn't see why it matters to them |
| Piece is interesting but hard to remember | Missing S (simple) - no single core idea |
| Piece is full of jargon and abstractions | Missing C (concrete) - no specific examples |
+
+ ## Stop conditions
+
+ Do not draft, or do not hand over, and say which condition fired.
+
+ 1. **Thesis scores 0 or 1.** Stop. Say the piece has no defensible position yet and ask for the
+ specific experience or result that would give it one.
+ 2. **No first-hand material at all**, meaning no story, no client pattern, no observation the
+ author personally has. The piece will be a summary of other people's ideas. Ask before writing.
+ 3. **The credibility slot is empty and the only fix is a statistic you do not have.** Never invent
+ the statistic. Switch to anti-authority or a testable credential, or write
+ `[NEEDS INPUT: the real figure]`.
+ 4. **The piece is a named executive byline and the user has not confirmed the executive holds this
+ view.** Mark it `[APPROVAL NEEDED: <name>]` and say it must not publish unconfirmed.
+ 5. **The thesis contradicts `knowledge/markets/positioning.md`.** Flag the conflict and ask which
+ one is meant to move. Publishing against your own position confuses the market twice.
+ 6. **The piece names a competitor and makes a factual claim about them** that is not in
+ `knowledge/markets/competitors.md`. Route to the user before it goes anywhere.
+
+ ## Warning signs in your own draft
+
+ Ranked worst first. The first four block handover.
+
+ 1. A statistic, quote or named example with no source. Blocker.
+ 2. A quote attributed to a real person who did not say it, in any form, including a paraphrase
+ presented as their words. Blocker, and the most damaging thing this skill can produce.
+ 3. A factual claim about a named competitor with no source. Blocker.
+ 4. An executive byline with no confirmation. Blocker.
+ 5. Section 3 restates section 2 in stronger language rather than contradicting it. The piece has
+ no unexpected element, and the reader will finish it agreeing with what they already thought.
+ 6. The story is a parable or a hypothetical rather than a scene that happened. Hypotheticals do
+ not carry credibility, and readers can tell.
+ 7. The close summarises rather than sharpening. If there is no line worth screenshotting, the
+ piece will not be shared regardless of how good the argument is.
+ 8. The new model has no name. Unnamed ideas do not travel.
+
+ ## Related skills
+
+ - `/linkedin-post` for the short version, and for the distribution post once this publishes
+ - `/newsletter-writer` when the piece is an issue of a recurring publication with a fixed spine
+ - `/content-writer` when the piece is shorter than 1200 words or is really a marketing asset
+ - `/copy-review` to grade and tighten a draft the user already wrote
+ - `/positioning-doc` when the thesis keeps conflicting with the company's stated position
+ - `/messaging-framework` when the new model needs to become language the whole company uses
+ - `/competitor-analyst` before any piece that makes factual claims about named competitors
+ - `/pr-pitch-writer` when the goal is placement in a trade publication rather than owned publishing
+ - `/content-repurposer` after publishing, to turn the piece into posts, a talk and a thread
+ - `/brand-context` first, whenever `knowledge/brand/voice.md` does not exist yet
+
+ ## What this skill cannot know
+
+ These are real limitations of this skill, and none can be settled from `knowledge/`. Ask the
+ user, or emit `[NEEDS INPUT: <what>]`.
+
+ 1. **Whether the named author actually believes this.** A byline is a public position taken by a
+ real person, and their view cannot be inferred from a brand voice file or a positioning doc.
+ 2. **Whether a customer or peer agreed to be quoted or identified.** An anti-authority quote is
+ the most persuasive element in the piece and the one most likely to have been given
+ informally, in a call, without permission to publish.
+ 3. **Whether a statistic from `knowledge/` is current, and whether its original source still says
+ what the note says it said.** Second-hand statistics drift, and this piece will be the third
+ hand.
+ 4. **Whether the publication target will accept the piece as written.** Trade publications have
+ house rules on promotion, sourcing and competitor mentions that are not visible from here.
+ 5. **Whether the position creates commercial or legal exposure** with a partner, an investor or a
+ regulator. Flag anything pointed, do not adjudicate it.