Copywriting

Copywriting is writing whose success is measured by what the reader does next: sign up, buy, reply, click, keep reading, remember the name. It is salesmanship in text. The craft lies less in clever phrasing than in understanding the reader so precisely that the copy says what they were already thinking — about their problem, their doubts, and what they'd need to believe to act.

Scope

Covers marketing and persuasive copy across channels: web pages, ads, email, product descriptions, store listings, social, sales collateral, launch announcements, naming and taglines.

Neighbours:

• ux-writing — interface text inside the product (buttons, errors, onboarding microcopy). Some overlap with CTAs and onboarding; ux-writing optimises for task completion, copywriting for persuasion.
• website-and-landing-pages — building the page; this skill writes what's on it.
• seo — making pages discoverable; this skill makes them convert once found. They interact (see below).
• experiment-design — designing rigorous A/B tests of copy variants.
• business-communication — internal and relationship communication.
• technical-writing — explaining; copy persuades. Developer-audience marketing often needs both.

Core principles

1. Research beats writing

Most weak copy is a research failure, not a writing failure. The best lines often come straight from customers: the words they use for their problem in reviews, support tickets, sales calls, forums, and interviews. Customers describe the pain more vividly than marketers do, and readers recognise their own language. Spend more time learning the audience than drafting.

2. One reader, one action

Write to a specific person in a specific situation and ask for one specific action. Copy aimed at "everyone" persuades no one, and a page with five equal CTAs splits attention until nobody does anything. Secondary actions can exist, but they must look secondary.

3. Meet the reader's awareness level

Readers arrive knowing different amounts. A useful lens (adapted from classic direct-response thinking) is a ladder of awareness:

Ad copy for a cold audience and an email to existing trial users about upgrading are opposites on this ladder. Starting with your product's features for an unaware reader, or with a long problem story for someone ready to buy, both fail.

4. Benefits over features — connected, not instead

A feature is what the product is or has. A benefit is what the reader gets. The reader cares about benefits, but they believe benefits because of features. The strongest copy chains them: feature → so what → so what, until you reach something the reader actually wants, then keep the feature nearby as the reason to believe.

Illustrative chain: "256-bit encryption" (feature) → your files can't be read if intercepted → you can share client documents without risking a breach → you keep client trust (and your professional obligations). The headline might lead with the outcome; the body cites the encryption as proof.

Two caveats. Expert buyers (engineers, procurement, specialists) often want specifications first — burying them under vague benefits reads as evasive. And benefit language without features becomes empty ("Unlock your potential").

5. Specific beats general

Specific claims are more credible and more memorable than general ones. "Saves time" is noise; "Closes your books in 2 days instead of 8" is a claim a reader can picture and evaluate. Specificity applies to numbers, customers ("used by 40 hospital pharmacies"), mechanisms ("we pre-match transactions using your bank's reference codes"), and outcomes. The constraint: every specific claim must be true and substantiated (see Ethics).

6. Clarity before cleverness

If a reader must decode the headline, most won't. Wordplay and abstraction work only when the meaning is still instantly clear, or when the brand is so well known that context does the work. A clear, specific headline usually outperforms a clever, vague one. Test clever ideas against a plain version rather than assuming.

7. The brand's voice, not yours

Voice is derived from the brand: its existing copy, guidelines, audience, and positioning. A children's education company, a cybersecurity vendor, and a streetwear label should not sound alike, and none should sound like the writer's personal style. Discover the voice first (see below); if none exists, propose one grounded in the audience and positioning, and keep it consistent.

8. Persuade honestly

Persuasion that relies on deception is both unethical and, in many jurisdictions, illegal — and it damages the brand once readers notice. Copy may emphasise, frame, and select; it may not mislead. Details under Ethics.

Process

Step 1: Brief

Clarify, or infer and state:

• Product/offer: what exactly is being sold or promoted, price, terms, what's genuinely different.
• Audience: who, their situation, the problem they have, what they've tried, their objections, awareness level.
• Action: the one thing the copy should get them to do, and what happens after.
• Channel and format: where it appears, length limits (ad character counts, subject-line truncation, store-listing limits), what surrounds it (images, form, price).
• Voice: brand guidelines, existing copy, words to use and avoid.
• Proof available: testimonials, data, case studies, certifications, press, customer counts — with permission to use them.
• Constraints: legal/regulatory review (finance, health, pharma, alcohol, children's products, and others have strict rules), required disclaimers, claims that are off-limits.
• Success metric: conversion rate, click-through, reply rate, signups, revenue per visitor.

Step 2: Research

• Mine customer language: reviews (yours and competitors'), support tickets, sales call notes, survey open-ends, community forums, interview transcripts. Collect exact phrases for problems, desired outcomes, and objections.
• Study competitors' positioning — not to copy it, but to find what's unclaimed and avoid sounding identical (see competitive-analysis).
• Verify the product's claims with whoever owns them: what's measured, under what conditions.

Step 3: Find the core message

Before writing lines, write the value proposition in plain words: for [audience] who [problem/desire], [product] [delivers outcome] because/unlike [reason to believe/difference]. This is a thinking tool, not copy to publish. If you can't complete it convincingly, the positioning is unclear — raise it with the user rather than papering over it with adjectives.

Step 4: Draft many headlines

The headline (or subject line, or ad's first line) decides whether anything else is read. Write many — twenty or more for important pieces — in different angles, then select. Angles to try:

• The outcome ("Close your books in 2 days")
• The problem ("Still reconciling transactions by hand?")
• The mechanism or what's new ("Bank-matched bookkeeping")
• The audience ("Accounting software for independent clinics")
• Proof ("How 400 clinics cut month-end from 8 days to 2")
• The objection removed ("Switch in an afternoon — we import everything")
• A specific question, story, or surprising fact

Good headlines are usually clear about who it's for or what they get, specific, and short enough for the channel. A subheading can carry the detail the headline can't.

Step 5: Structure the body

For longer copy (landing pages, sales emails), a reliable order is: hook → problem or desire (in the reader's words) → the solution and how it works → benefits backed by features → proof → objections answered → offer → CTA. Classic formulas (AIDA: attention, interest, desire, action; PAS: problem, agitate, solve; before–after–bridge) are variations on this — use them as checklists, not scripts. Agitation, in particular, should stay truthful and proportionate; manufactured anxiety backfires.

Length should follow the decision's weight and the reader's awareness: an impulse purchase or a known brand needs little; a considered, expensive, or risky decision needs enough copy to answer every serious question. "Long copy doesn't work" and "short copy always wins" are both myths; what fails is copy that's longer than the reader's interest or shorter than their doubts.

Step 6: Add proof

Readers discount claims they can't verify. Strengthen them with:

• Specific testimonials with name, role, and context (with permission), focused on a concrete outcome rather than generic praise.
• Data: customer counts, results, ratings — with source and date.
• Case studies: before, intervention, after.
• Authority: certifications, standards compliance, credible press, known customers' logos (with permission).
• Demonstration: screenshots, video, a free trial, a live demo.
• Risk reversal: guarantees, free trials, easy cancellation — but only if genuinely offered.

Place proof next to the claim it supports, not quarantined in a "testimonials" block.

Step 7: Write the CTA

• Use a verb describing what the reader gets or does: "Start my free trial", "Get the pricing guide", "See plans" — not "Submit" or "Click here".
• Reduce perceived risk near the CTA: "No credit card required", "Cancel anytime" — only if true.
• Make the next step's commitment level match the reader's readiness; a cold reader may "See how it works" before "Book a demo".
• Make the primary CTA visually and verbally dominant; repeat it at natural decision points on long pages.
• Ensure CTA copy matches what actually happens next. If the button says "Get instant access" and leads to a sales form, trust breaks.

Step 8: Edit ruthlessly

• Cut throat-clearing and empty superlatives ("world-class", "cutting-edge", "seamless", "revolutionary"). If a claim could be made by any competitor, it differentiates nothing.
• Replace abstract words with concrete ones.
• Prefer "you" over "we" — the copy is about the reader.
• Read aloud: copy that's hard to say is hard to read.
• Check scannability: a reader skimming only headings, bold text, and CTAs should still get the message.
• See writing-review-and-editing for a full editing pass, and fact-checking for claims.

Step 9: Produce variants and plan the test

Deliver the recommended version plus a small number of genuinely different alternatives — different angles, not synonyms — and say what each tests. See Testing below.

Discovering the brand's voice

1. Read existing copy across channels: website, emails, ads, social, product UI, support macros. Note formality, sentence length, humour (and its kind), person (we/you/they), use of jargon, punctuation habits, how claims are framed.
2. Find guidelines: brand book, voice and tone guide, messaging framework, glossary of approved and banned terms, legal guidance on claims.
3. Identify the audience the voice was built for; voice that fits a consumer audience may need adjusting for enterprise buyers within the same brand — usually as a shift in tone, not a different voice.
4. Note mandatory elements: trademark usage, product name capitalisation, required disclaimers.
5. Describe the voice as contrasts if you must articulate it: "confident, not arrogant; plain-spoken, not casual". This is more usable than adjective lists.

If the brand has no voice yet, propose a short description derived from audience and positioning and show it in the copy; let the user confirm before producing large volumes.

Decision guidance

Channel considerations

Always check current character limits and platform policies for the specific channel, as they change.

Audience-specific adjustments

• Business buyers (B2B): several people decide — the user, the budget holder, IT/security, procurement. Copy may need to equip a champion to persuade others: ROI framing, security and compliance facts, integration details, implementation effort. Proof from similar companies (industry, size) matters more than general popularity.
• Developers and technical audiences: sceptical of marketing language; respond to specifics, code samples, honest limitations, transparent pricing, and fast paths to trying the thing. Show, don't hype. Often overlaps with technical-writing.
• Consumers: emotion, identity, convenience, and social proof weigh heavily; decisions may be fast. Clarity about price, delivery, and returns removes friction.
• Existing customers: they know you; skip the introduction. Lead with what's new for them and what to do.
• Non-native readers and international markets: plain language, no idioms; localise (not just translate) offers, examples, proof, and cultural references; check that names and slogans don't carry unintended meanings.

Naming and taglines

• Generate many candidates across types: descriptive (says what it does), suggestive (evokes a benefit), invented/abstract (distinctive, needs marketing to gain meaning), and founder/place names.
• Evaluate against: memorability, pronounceability, spelling from hearing it, distinctiveness from competitors, fit with the brand, unintended meanings across target languages, and availability (domains, social handles).
• Trademark clearance is a legal matter — flag it, never declare a name "available" on the basis of a web search.
• Taglines should say something the brand can own; "Quality you can trust" could belong to anyone.

Handling feedback on copy

Stakeholders often react to copy with taste-based comments ("make it pop", "I don't like this word"). Translate them into the underlying concern — is the problem clarity, credibility, tone, or accuracy? — and address that. Reference the brief and evidence (customer language, test results) to resolve disagreements, and test when opinions are split and traffic allows. Legal and compliance edits are not negotiable taste; work within them.

When the product isn't ready for the copy

Sometimes the copy can't be strong because the substance isn't there: no clear differentiation, no proof, an offer that's worse than competitors'. Don't cover the gap with superlatives. Instead:

• Say what's missing ("there's no evidence yet for the speed claim; a benchmark or two customer quotes would fix that").
• Lead with what is true and distinctive, however modest — specificity about a narrow strength beats vague claims about broad ones.
• Suggest cheap ways to create proof: a pilot with measured results, a free trial, a guarantee the business can honour, a transparent comparison.
• Narrow the audience to the segment for whom the product genuinely is the best choice.

Copy amplifies positioning; it cannot replace it.

Copy and SEO

Search-oriented pages must satisfy both search intent and conversion. Write for the reader first; incorporate the language searchers actually use (which is often customer language anyway) naturally in titles, headings, and body. Don't stuff keywords — it reads as spam to humans and to search engines. See seo.

Testing variants

• Test big differences first. Angle, offer, headline concept, page structure, and proof type move results more than synonyms or button colours.
• One hypothesis per test, stated in advance ("Leading with the time-saving outcome will lift trial starts over leading with the integration list, because interviews show time is the top pain").
• Enough traffic: small sites can't detect small differences; with low traffic, test bold variants or use qualitative methods (user tests, five-second tests, customer interviews) instead. See experiment-design and statistical-analysis for sample size and stopping rules.
• Measure the right outcome: clicks are easy to inflate with curiosity-bait; measure the downstream action (purchase, activation, retention) where possible.
• Watch for novelty effects and segment differences; a winner for new visitors may lose for returning ones.
• Record results and the reasoning, so learnings accumulate instead of being re-tested.

Ethics and legal guardrails

Non-negotiable:

• No false claims. Never invent statistics, testimonials, reviews, customer logos, awards, press mentions, or certifications. If you need a proof point that doesn't exist, leave a clearly marked placeholder and tell the user what evidence is needed.
• Substantiate objective claims. "Fastest", "#1", "clinically proven", "saves 40%" need evidence the advertiser can produce. Many regulators (e.g. consumer-protection and advertising-standards bodies) require substantiation before the claim is made.
• Comparative claims about competitors must be accurate, current, and fair.
• Testimonials must reflect real, typical experiences or clearly disclose atypical results; disclose material connections (paid endorsements, free products, affiliate relationships).
• No fake urgency or scarcity. Countdown timers that reset and "only 2 left" when false are deceptive.
• No dark patterns. Confirmshaming ("No thanks, I don't like saving money"), hidden costs, pre-checked add-ons, disguised ads, and obstructed cancellation erode trust and increasingly violate law.
• Regulated categories (financial products, health and medical, supplements, pharmaceuticals, alcohol, gambling, children, employment, housing, political ads) have specific rules on claims, disclaimers, and targeting. Flag these and recommend legal or compliance review.
• Price and terms must be clear: total cost, recurring billing, trial conversion, cancellation terms.
• Respect consent and privacy law in email and messaging (opt-in requirements, unsubscribe, sender identification).
• Don't target vulnerability exploitatively — e.g. playing on health anxiety, financial desperation, or addiction.

When a user asks for something that crosses these lines, explain the risk briefly and offer an honest alternative that still persuades. Honest specificity is usually more persuasive than inflated claims anyway.

Pitfalls

• Writing before researching. Produces generic copy in the writer's words rather than the customer's.
• Feature lists without meaning. "AI-powered, cloud-native, real-time" — the reader doesn't know why they should care.
• Benefits without substance. "Transform your workflow" — unbelievable because unsupported.
• Talking about the company, not the reader. "We are a leading provider of…" Count "we" vs. "you".
• Superlatives and buzzwords any competitor could claim.
• Clever but unclear headlines.
• Mismatched message: the ad promises one thing, the page talks about another; the reader bounces.
• Too many CTAs or a vague CTA ("Learn more" everywhere).
• Ignoring objections: price, switching cost, trust, "will it work for me?" left unanswered.
• Inventing proof, or using testimonials and logos without permission.
• Voice drift: copy that sounds like a generic marketing template rather than the brand.
• Testing trivia: endless button-colour tests while the offer and headline go untested; declaring winners on insufficient data.
• Exceeding channel limits so the key words are truncated.

Quality bar

• The audience, their awareness level, and the single desired action are explicit (in the brief or your report).
• The headline is clear, specific, and relevant to that audience; a skimmer gets the message from headings and CTA alone.
• Benefits are connected to features or mechanisms that make them believable.
• Every factual claim is true and substantiated, or flagged as needing evidence; no invented proof.
• Main objections are addressed.
• The CTA is specific, matches what happens next, and dominates secondary actions.
• Voice matches the brand's existing copy or guidelines.
• Channel constraints (length, truncation, disclaimers, policies) are met.
• No dark patterns, fake urgency, or deceptive framing; regulated-category concerns are flagged.
• Variants differ meaningfully and each has a stated hypothesis.
• Read aloud once; it sounds like a person talking to the reader.

When delivering, explain the angle you chose and why, list assumptions about audience and proof, and mark anything that needs fact-checking or legal review.

Illustrative example: rewriting a hero section

Illustrative only; product and figures are hypothetical.

Original: "The Future of Scheduling Is Here" / "Our innovative, AI-powered platform streamlines your operations with seamless integrations." / [Learn More]

Diagnosis: no audience, no specific outcome, buzzwords, unsubstantiated "innovative", vague CTA. Research (support tickets, interviews) shows the buyers are dental practice managers whose main pain is no-shows and the phone time spent rebooking.

Rewrite (recommended): "Fill no-show gaps before the chair goes empty" / "When a patient cancels, [Product] texts your waitlist and books the first reply — automatically. Practices using it recover an average of [verified figure] appointments a month." / [Start a 14-day free trial] · "Works with [named practice software]. No card required."

Variant B (problem angle, for a colder audience): "Spending your mornings calling patients to fill cancellations?" — tests whether naming the pain beats naming the outcome.

Notes to the user: the average-recovery figure must come from real customer data (placeholder until verified); confirm the trial terms and integration list are accurate; the CTA assumes trial signup is the primary goal.