Copywriting: Write and Diagnose Persuasive Copy

This skill turns a century of direct-response and brand craft, from Claude Hopkins to Alex Hormozi, into one repeatable operating procedure for writing copy that converts and for diagnosing copy that doesn't. The depth lives in references/; this file is the decision logic that tells you what to do and when to go read it.

The one idea everything serves

Eugene Schwartz, Breakthrough Advertising (1966): "Copy cannot create desire for a product. It can only take the hopes, dreams, fears and desires that already exist in the hearts of millions of people, and focus those already existing desires onto a particular product."

That reframes the whole job from convince to channel. You are not manufacturing want out of nothing. You are locating the want that already exists, then redirecting it onto this product with specific, believable claims. Every technique in this skill is a tactic in service of that one strategy.

The practical consequence is the thing that separates masters from amateurs: diagnose before you write. Amateurs reach for a formula and start typing. Masters first figure out who they are talking to, what that person already believes and wants, and how many similar pitches they've already tuned out, and let that determine everything downstream. A brilliant sentence aimed at the wrong awareness stage still fails.

Two modes

This skill runs the same engine in two directions:

• Write mode: produce new copy. Walk the procedure forward (diagnose → idea → structure → hook → body → offer → edit).
• Diagnose mode: the user pastes existing copy and asks why it's flat / not converting / too salesy, or asks for a critique or rewrite. Run the procedure in reverse: grade the copy against each step, find where it leaks, and fix the highest-leverage leak first (almost always the headline/lead or the offer, not the prose). Then deliver a tightened rewrite, not just notes.

Most real requests are one of these. Some are both ("here's my draft, make it actually sell"): diagnose, then rewrite.

Step 0: Lead with the deliverable; keep the reasoning proportional and behind it

Always diagnose internally, but remember the prospect's job and the user's job differ. The user asked for copy, so give them the copy first. Default to deliverable-first: the headline / email / page they can paste, then a short "why this works" after it. Opening with a block of strategy forces the reader to scroll past your homework to reach the thing they came for, exactly the warm-up the manual tells you to cut. Front-load the diagnosis only when the user is explicitly asking for the thinking (positioning help, a teardown, "what's my angle?"), or when you're co-developing a big sales page where the strategy is the point.

Scale the visible reasoning to the ask, too: one subject line earns a line of rationale or none; a sales page earns a short paragraph. Over-explaining a small ask is its own kind of friction.

Keep the strategy vocabulary backstage. Awareness stages, sophistication, mechanisms, the Life-Force 8, the legends' names. Those are your tools for thinking, not the reader's. The copy you hand over, and even the "why it works" note, should read like a sharp human wrote it, never like a marketing textbook citing its sources.

When you genuinely lack the information to diagnose (you don't know the audience, the product, or the offer), prefer one or two high-leverage questions over a dozen. Or, better, state your assumptions explicitly and write to the most probable version of the prospect, flagging where real customer language would sharpen it. Don't stall a writing task behind an interrogation; the field's own discipline is "a draft beats a blank page," and you can mark assumptions for the user to correct.

The operating procedure

Nine steps. Each has a trigger and an output that feeds the next. Hybridize and compress freely: on a warm, known, ready-to-buy audience you can skip straight to the offer; on a cold, jaded one you'll do all nine.

1. Mine the language (Voice of Customer). The best copy is found, not written (Joanna Wiebe). Before inventing claims, harvest the market's actual words, from reviews, forums, Reddit, support tickets, sales-call transcripts, anything the user can give you. Pull the verbatim pains, desired outcomes, and objections. Output: you can state the prospect's pain and dream in their words, not yours. If you have no source material, say so and write to the most likely language, then flag it. → deeper method in references/mechanics.md.

2. Diagnose the two axes. This is the heart of the system. Pin the prospect on:

• Awareness (how much do they know?): Unaware → Problem-aware → Solution-aware → Product-aware → Most-aware. This sets your lead and how long before you mention the product.
• Sophistication (how many similar claims have they already heard?): Stage 1 first-to-market → 2 outbid the claim → 3 introduce a mechanism → 4 elaborate the mechanism → 5 pivot to identity/brand.

Use this quick map; read references/diagnostics.md for the full model, decision rules, and worked examples:

If you're unsure, assume the harder case (lower awareness and higher sophistication) and lead with story/identity plus a mechanism. It costs a little length and rarely backfires; the reverse (pitching above the prospect's stage) silently kills response.

3. Find the Big Idea + unique mechanism. One dominant idea per message: "one message, well developed, just has more impact than ads overloaded with competing ideas." Run Ogilvy's five tests (did it make you gasp; do you wish you'd thought of it; is it unique; does it fit the strategy; could it run for years). If the bare claim has gone stale (Sophistication 3+), give it a unique mechanism, a named how that makes the promise newly credible. → references/mechanics.md.

4. Pick (and plan to hybridize) a structure. Choose by awareness and temperature, don't cargo-cult a formula:

• Low awareness / cold → Star-Story-Solution, PASTOR, or the 12-step VSL spine
• Problem-aware → PAS (Problem-Agitate-Solution)
• Solution/Product-aware → AIDA, the 4 Ps, FAB
• Most-aware → lead with the offer / the Stack

Long pages legitimately blend these (PAS to open, FAB/4 Ps in the body, AIDA/Stack to close). → full catalog + when-to-use in references/formulas.md.

5. Engineer the hook. The headline is most of the job (Ogilvy: ~80% of the dollar; Caples: 50 to 90% of the effect). Write 10-20 headline options, not one. Grade each against the 4 U's (Urgent, Unique, Useful, Ultra-specific) and the "so what?" test. The best one enters the conversation already in the reader's head (Collier) and opens a curiosity loop. The first sentence should be short, almost incomplete, so it's easier to read than to skip (Sugarman's slippery slide). → references/mechanics.md for curiosity mechanics.

6. Dimensionalize benefits + stack proof. Convert every feature → benefit → vivid felt end-state ("your golf buddies' jaws drop when your ball flies 40 yards past theirs"), not just "improves performance." Attach a proof element to every claim, and never make a claim bigger than your proof (Bencivenga). Pre-empt the top 3-5 objections inline. Raise the obvious flaw yourself to earn trust (Sugarman's honesty trigger). → references/mechanics.md.

7. Construct the offer. Often higher-leverage than the prose. Build the value stack and reveal it before the price so price anchors to the whole; justify the price with a rationale; add bonuses rather than discounting the core; choose a risk-reversing guarantee; apply only honest urgency/scarcity. Pressure-test with Hormozi's Value Equation, value = (dream outcome × perceived likelihood) ÷ (time delay × effort), and Suby's "would this offer keep you up at night?" → references/offers.md.

8. Edit for slide and rhythm. Insert bucket brigades ("Here's the deal:", "But it gets better:") wherever attention could drop. Cut friction words and jargon. Vary sentence length deliberately: mix short, medium, and long so the rhythm has music instead of droning (Provost); favor short, but let an occasional long sentence build. Use "you" relentlessly. Read it aloud: wherever you stumble speaking it, the reader stumbles reading it (Halbert), and the ear catches uniform rhythm the eye misses. Apply the grunt test: could a stranger grasp what it is, how it helps, and how to buy in ~5 seconds? Then ship.

Sound human; cut the machine tells. Generic, template-flavored copy reads as committee-written and untrustworthy, and untrustworthy copy doesn't convert. Run a dedicated subtractive pass to strip the tells that make prose feel machine-assembled: the autopilot vocabulary (delve, leverage, robust, seamless, elevate, unlock, testament, vibrant, "it's important to note," "in today's fast-paced world"), the "it's not just X, it's Y" construction, the reflexive rule of three, "from X to Y" false ranges, pseudo-analytic "-ing tails" ("…underscoring its significance"), tidy "in conclusion" wrap-ups, significance inflation, and vague attribution ("experts say"). Replace each with how a sharp human would actually say it: plain verbs, a real stance, a concrete specific. This is the same craft the legends always demanded (Halbert's read-aloud, Ogilvy's respect for the reader), not a detector trick, so never use the quality-destroying gimmicks (typos, invisible characters, thesaurus-swapping, "humanizer" tools); they ruin readability and have no place in copy meant to sell. → full checklist in references/mechanics.md.

Punctuation discipline: go light on the em dash. Stacked em dashes are the clearest punctuation tell of machine-written copy, and overuse makes prose feel breathless and assembled by formula. Cap them at about one per paragraph, and reach for one only when no other mark fits. Default to a period (split into two sentences), a comma (a quick aside), a colon (a setup and its payoff), or parentheses (a true aside). The em dash is a real rhythm tool, so don't ban it outright; when you cut one, keep the beat it carried. On your final pass, scan for em dashes and convert every one the sentence does not actually need.

9. Recommend what to test. Judge by response, not applause (Hopkins, Caples, Wiebe). Name the one or two highest-leverage things to A/B first: almost always the headline and the offer. If you can, hand the user a second headline/lead variant to test against the primary.

What to hand back

Write mode:

1. The copy itself, first: formatted for its channel and ready to paste. This is what the user came for; don't bury it under preamble.
2. 2-3 alternative headlines / subject lines: the highest-leverage element to vary, so hand over options rather than one guess.
3. A short "why it works" note after the copy (scaled per Step 0): who you pictured and where they are, the core desire and the Big Idea/mechanism you built on, and what to test first (almost always the headline or the offer). A few lines for a small ask; a tight paragraph for a sales page. The reasoning should be available without the user having to wade through it to reach the copy.

Place that "why it works" block before the copy only when the user asked for strategy rather than finished copy.

Diagnose mode:

1. The single biggest leak, in a line or two: the one thing costing the most conversions, followed immediately by
2. The rewrite of at least the headline/lead and any section you flagged. Show the fix before you explain it; a pasted-in improvement is worth more than a paragraph about it.
3. The prioritized critique after the rewrite: the few fixes that matter most, ordered by leverage (headline/offer before line-edits), each with a concrete before → after, ending with what to test first. Resist listing every flaw: three ranked fixes beat ten unranked ones, because the user can only act on so much at once.

Guardrails: persuasion, not manipulation

The greats are nearly unanimous that honesty is the most powerful long-term trigger (Sugarman ranks it #1), and these lines protect both the reader and the result:

• Never a claim bigger than your proof. Specificity and demonstration build belief; unbacked superlatives invite skepticism. If proof is thin, make a smaller, truer claim. It converts better anyway.
• Honest urgency and scarcity only. Real deadlines and real limits, disclosed plainly. Fabricated countdowns and false "last chance" convert short-term and erode trust (and create legal exposure in regulated verticals) long-term.
• Don't manufacture desire or fake emotion: channel what's real. Don't invent testimonials, statistics, certifications, or outcomes. If the user hasn't supplied proof, use a placeholder and tell them to insert the real thing.
• The customer is the hero, not the brand (StoryBrand). Name the internal and philosophical problem, not just the external one, but don't exploit fear or insecurity to do it.
• Formulas are thinking tools, not laws. No fixed sequence of effects is universal (the academic record is clear on this); attention, interest, desire and action fire in parallel. Use structures to organize the argument, then trust judgment over template.

Reference map: read these as needed

Pull the relevant file into context when a step calls for depth; you don't need them all every time.

• references/diagnostics.md. Schwartz's 5 Awareness Stages and 5 Sophistication Stages in full, the decision rules, the UPSYD variant, and how the two axes combine to set your angle. Read this whenever you're unsure how to lead.
• references/formulas.md. Every structure (AIDA, PAS, PASTOR, BAB, 4 Ps, 4 U's, FAB, QUEST, Star-Story-Solution, the 12-step VSL spine), what each is for, and how to hybridize. Read when choosing or combining a structure.
• references/mechanics.md. The transferable moves: Big Idea, features→dimensionalized benefits, the curiosity engine (Zeigarnik, open loops, curiosity gap, bucket brigades), specificity, emotional drivers (Cialdini's 7, Whitman's Life-Force 8), objection pre-emption, proof, voice/rhythm/readability, the sound-human / cut-the-machine-tells editing pass, and the Voice-of-Customer research method. Read when writing or editing the body.
• references/offers.md. Offer architecture: the Value Equation, the Grand Slam / Godfather offer, the value Stack, the four guarantee types, pricing/anchoring, and ethical urgency. Read when the offer (not the prose) is the lever.
• references/legends.md. The canon: each legend's signature contribution distilled to a reusable principle, plus the caveats (formula skepticism, source-quality notes, the ethics of urgency). Read for background, for the "why," or to channel a specific master's approach.