Copywriting

Writes persuasive marketing copy across formats, covering tone, structure, clarity, and the discipline of selling benefits rather than features.

Benefits, not features

• The so-what ladder: take a feature, ask "so what?" 3-5 times until you hit a felt need (safety, status, freedom, belonging, relief). The #1000 stops at the surface benefit ("saves time"); the #1 reaches "home by 6 for dinner".
• Test: a feature answers "what it does"; a benefit answers "what the reader gets or feels". If the reader could reply "so what?", it is still a feature.
• B2B keeps features, but as proof, not lead: the buyer compares specs, so the spec table earns its place, but the headline and first line are still the benefit. Lead benefit, hold the feature as evidence.
• Concrete nouns carry benefits implicitly: "8 hours of sleep" beats "better rest" because the reader pictures it. Abstract benefits make the reader do the work you skipped.
• An adjective is an unquantified feature: "fast" -> "loads in under a second"; "reliable" -> "99.9% uptime"; "powerful" -> "400 hp". If you can't quantify it, you don't know it well enough to sell it.
• Translate your feature list into a benefit list before writing a word of copy; the feature list is raw material, not copy.

Headlines

• Most of the result is here: roughly 5x more readers see the headline than the body (long-standing direct-response benchmark). The headline is the ad; the body is the detail.
• Write 5-10 before picking: the first 3 are the clichés every competitor writes; usable ones start around #5. One headline is a guess; eight is a choice.
• Specificity beats cleverness: "Save $1,847 on taxes" outpulls "Save money on taxes". Numbers, proper nouns, and exact quantities read as real; generalities read as filler.
• The "would I read the next line?" test on every headline. If no, the copy is dead before the body starts.
• Recurring structures (not novelty): how-to, question, news, promise, list, command, story-curious. Match structure to reader intent, not to taste.
• Length by medium: print grabs in 5-7 words; digital rewards specificity up to ~60-70 chars before truncation, but the first 3 words carry mobile. Search and feeds reward the longer specific headline, not the short clever one.
• Subheads carry the scanner: a well-built page reads by headline + subhead + CTA alone. Write subheads as a parallel set, not as afterthoughts.

Clarity over cleverness

• Read it aloud; where you stumble, the reader stops. The #1 edits by ear, the #1000 edits by eye.
• Reading level: grade 6-8 for consumer copy, up to 9-10 for B2B technical. Higher grade = lower comprehension, not higher status.
• One idea per sentence; two ideas = a paragraph. Long sentences only to vary rhythm right after a punchy short one.
• Concrete beats abstract, Anglo-Saxon beats Latinate: "truck" not "vehicle", "sweat" not "perspiration", "use" not "utilize". The plain word is shorter and clearer.
• Active voice; passive hides the actor and the stakes. "Mistakes were made" is a confession dressed as copy.
• Cut on sight: "very", "really", "just", "actually", "simply", "that" (often), "in order to" -> "to". Each is a word adding zero meaning.
• Corporate filler signals you had nothing to say: "leverage", "synergy", "world-class", "cutting-edge", "best-in-class", "solution". Delete or replace with the specific.

Structure: lead, body, close

• The lead earns the next sentence; that sentence earns the next. Each line is a hook for the line after it, not a standalone claim.
• Lead with the strongest benefit or the most pointed pain, not the company history. The reader's first question is "what's in it for me", answered in the first 2-3 lines or never.
• PAS (Problem, Agitate, Solution) for cold readers who don't yet feel the problem; AIDA assumes an Interest that cold traffic doesn't have. Match the frame to where the reader starts.
• The body earns the CTA; the CTA is the receipt, not the surprise. A weak body makes even a great CTA feel pushy.
• One CTA per page or email; two split attention and cut response. The exception is a secondary low-commitment option ("or talk to sales") when buyers genuinely differ, as in B2B.
• Proof goes nearest the claim it supports, not in a heap at the bottom. A testimonial three screens from its claim proves nothing in the reader's memory.

Calls to action

• Specific verb beats generic: "Get my free quote" > "Click here" > "Learn more". The CTA names the next action and its reward.
• First-person can outperform second-person: "Start my plan" vs "Start your plan" tests cheaply and often wins in direct response; the reader owns the action.
• Friction words carry cost: "Submit", "Enter", "Buy" feel like work; "Get", "Start", "Send", "Claim" feel like movement. Pick verbs that move toward the benefit.
• Placement by intent: CTA above the fold for warm traffic; after the argument for cold. A CTA above a weak argument converts worse than no CTA, because it asks before it earns.
• The button is copy: it answers "what happens when I click". "Start 14-day trial" beats "Submit"; "Send me the report" beats "Download".

Tone and voice

• Voice is consistent across surfaces; tone shifts with context. The refund email and the launch email share a voice, not a tone.
• Match the reader's register, not yours: a CFO and a first-time founder need different vocabularies for the same product. Write to the person who decides, at their level.
• Write to one person, not an audience: "you", never "users" or "customers" in the copy itself. "Our users love it" reads as a press release; "you'll feel it on day one" reads as a letter.
• The pronoun test: count "you" vs "we". More "we" than "you" = self-centered copy; aim for ~2-3x "you".
• Wit works only when the audience gets the joke and the joke carries the benefit. Misfired wit costs more than plainness, and you won't see it miss.

Editing: the cut

• First draft says everything; the edit cuts everything that isn't earning its spot. The final is often 40-60% shorter than the first draft (range, not rule).
• The cut test: does this word change the reader's decision? No -> cut it. Mercy on the reader, not on your effort.
• Read with every other paragraph deleted; if it still works, those paragraphs were decoration.
• Adjectives first: replace with specifics. "Powerful engine" -> "400 hp"; "easy" -> "set up in 4 minutes".
• Read backwards, last sentence first: it breaks the forward flow that hides awkwardness and repetition.
• Reading-level tools are a blunt check, not a goal: use them to catch grade-level spikes, then fix the sentence, not the score.

Persuasion, used honestly

• Specificity is proof: "trusted by 2,847 teams" beats "trusted by thousands". Round numbers read as invented; odd, specific numbers read as counted.
• Scarcity only when true: a fabricated countdown converts once and burns the list. Real scarcity (limited seats, a closing cohort) is the strongest honest lever; fake scarcity is the fastest way to lose trust.
• Loss framing ("don't lose $X") can outperform gain framing on problems the reader already feels; gain framing wins for aspiration and new markets. Test both; the frontier is whether the pain is felt or imagined.
• Specific testimonials beat generic: "Cut reporting from 3 days to 2 hours" (with name, role, company) > "Great product!". A testimonial without a metric is a quote, not proof.
• Social proof proves popularity, not fit: a logo strip says "others use this", never "this is right for you". Pair it with a fit signal.

Format by surface

• Landing page: one promise, one audience, one CTA. Multiple audiences or offers on one page dilute all of them; split the page or split the traffic.
• Email: subject line is the headline, preheader is the subhead. Subject earns the open; first line earns the read; CTA earns the click. Three jobs, three lines.
• Email subject: 30-50 chars before mobile truncation; first 3-5 words carry it. Lead with benefit or curiosity, not the brand name.
• Ads: the headline carries most of the weight; image and first body line do the rest. Long-form copy works in direct response (long sales pages) and fails in interruption media (banners, feeds).
• Product page: benefits above the fold; features and proof below. The spec table serves the comparison shopper, not the first-time visitor; order the page for both.
• B2B copy must survive being forwarded: the user, the buyer, and finance each read only their section. Make every section stand alone, or it dies in the forward.
• Social is native, not resized: a LinkedIn post and a tweet are different formats with different rhythms. Same message, different copy.

Testing copy

• One variable per test: changing headline and CTA together teaches nothing about what worked.
• Sample size: a "winner" on 200 visitors is noise; aim for ~100+ conversions per variant before trusting direction, and it still depends on effect size and baseline.
• Novelty effect: a new variant often spikes then regresses; wait a full week or a full cycle before calling it.
• Peeking inflates false positives: set the test duration upfront and decide at the end, not the first day you're ahead.
• Real lift is modest: a 10-30% lift on a headline is a strong result; a 2-3x claim is almost always noise or a broken control. Calibrate expectations to the medium.
• Read losing variants: the copy that lost tells you what the audience rejected. It is data, not waste.

Situations

Situation	Play
Reader could reply "so what?"	Still a feature. Run the so-what ladder 3-5 levels to the felt need, then write that.
Cold traffic landing page	Lead with pointed pain or strongest benefit in 2-3 lines; CTA after the argument, not above.
Warm or retargeting traffic	CTA above the fold; they know the problem, give them the next step fast.
High-price B2B offer	Long copy, benefit lead, features as proof, secondary CTA ("talk to sales") for the buyer who won't self-serve.
Email open rate dropped	First 3-5 words of the subject carry mobile; lead with benefit or curiosity, not the brand; keep under ~50 chars.
Two CTAs splitting response	Cut to one. Keep the second only if buyers genuinely differ (B2B self-serve vs sales).
A/B test "won" on Monday	Peeking plus novelty. Set duration upfront, need ~100+ conversions per variant, decide at the end of a full cycle.
Copy reads generic ("powerful, easy, reliable")	Replace every adjective with a specific number or concrete noun; if you can't, you don't know the product.
Ad headline not pulling	Write 5-10 more; the first 3 are the competition's clichés; lead with a specific number or proper noun.
Clever line, flat conversion	Wit misfired. Trade the clever line for the plain benefit and test; plain wins when the stakes are high.

Where camps disagree

• Long vs short copy: long wins when the reader is already interested and the offer needs explaining (high price, complexity, risk); short wins in interruption and low-friction offers. The frontier is reader intent plus offer complexity, not ideology.
• Direct response vs brand: DR optimizes the next click, brand optimizes the decade. Pure-DR burns a list for short-term gain; pure-brand never asks for the sale. The #1 sequences them, roughly 60/40 brand/activation as an average across cases (IPA), conditional on maturity and category; applied as a constant, it reads only the summary.
• Clever vs plain: wit wins when the audience gets the joke and the joke carries the benefit; plain wins for broad audiences or high stakes (health, money, fear). The cost of misfired wit exceeds the cost of plainness, and you can't always see it miss.
• Fear vs aspiration: loss framing for problems the reader already feels; gain framing for futures the reader wants. A mismatch reads as either preachy or empty.
• Story vs straight pitch: story earns attention the pitch can't buy, in cold traffic and saturated markets; straight pitch wins for warm traffic who already want the answer. A story that never reaches the offer is entertainment, not copy.
• Features in B2B: the spec sheet matters because the buyer compares, but the headline and lead are still benefits. The frontier: features as proof after the benefit, never as the lead.

Related Skills

Install with clawhub install <slug> if the user confirms:

• branding - voice and positioning this copy expresses
• content-marketing - the owned-channel strategy copy serves
• landing-page - the surface where most persuasive copy is tested
• email-marketing - the format with the most copy-driven metrics
• ads - paid acquisition copy; overlap on headline and CTA craft