TL;DR
Launch day needs six pieces of copy at once: announcement, landing hero, email, press blurb, 60-second demo script, and FAQ. Write one shared brief, then run six copy-paste prompts to draft all six in parallel, then do a hard human pass. Realistic budget: about 90 minutes for usable drafts. The model you point these prompts at matters less than the brief and the human edit — but for tone and low-edit first drafts, Claude (Opus 4.7 / Sonnet 4.6) tends to need the least rewriting; for fast variant generation, ChatGPT (GPT-5.5) is quick. Every length rule below is a verified 2026 number, not a vibe.
Who this is for
Founders and indie makers shipping a product, feature, or campaign without a copywriter on staff — the person who realizes three days out that they need six assets and have no template.
When to reach for it
Three to five days before launch, when you suddenly need six pieces of copy and have two days to write them. If the launch concept itself is still fuzzy, run an AI campaign ideation workflow first to go from a one-paragraph brief to scored concepts before you write a single headline.
The length rules these prompts enforce (June 2026)
The prompts below hard-code current best-practice limits so the AI does not over-write. Sources are linked in the FAQ.
| Asset | Target length | Why this number |
|---|---|---|
| Landing hero headline | Under 8 words / ~44 characters | High-converting H1s in 2026 studies sit under 8 words; longer reads as jargon |
| Email subject | 30–50 characters | Optimal open-rate band; key words must survive the first ~40 chars |
| Email preheader | ≤ 80 characters | Acts as a second headline; should complete the subject’s thought |
| Press release | 300–500 words (~400) | 68% of journalists prefer releases under 400 words |
| Demo voiceover | ≤ 150 words | ~150 words is roughly 60 seconds of natural narration |
| FAQ answer | ≤ 2 sentences | Skimmable; one direct answer plus one reason |
Step by step
Every step ships a copy-paste prompt. Replace the [...] placeholders with your real info, then send to ChatGPT, Claude, or Gemini. Keep the same chat thread so the model carries your brief through all six pieces.
-
One brief, shared across all six pieces. Have the AI read it back before generating anything:
You are a senior launch-day copywriter. I will give you a brief; all 6 pieces below derive from it. Product (1 sentence): [function + key differentiation] Target audience: [role + what they use today + current pain] 3 story points: 1. [why now: origin / trigger] 2. [how the product solves it: the key moment] 3. [a concrete user outcome: number / quote] Launch goal (pick 1): signups / sales / press / demand validation Voice: [e.g. "direct, no hyperbole, few adjectives"] Forbidden phrases: [e.g. "Introducing / Excited to announce / revolutionary / game-changer"] Read the brief back to me so I can confirm you understood it. Do not generate any copy yet. -
Announcement (3 lengths, same spine):
Based on the brief, write the announcement in 3 lengths (same story spine, length varies): - X / LinkedIn: under 80 words, first line must work as a standalone hook - Blog: 200 words, structure = pain 1 graf / solution 2 grafs / CTA 1 line - Newsletter: 600 words, include 1 concrete-use-case anecdote Avg sentence length <=15 words. No emoji bullets. No forbidden phrases. -
Landing hero (5 angles):
Based on the brief, write 5 landing hero candidates. Each contains: - Headline (under 8 words, ~44 characters max) - Subhead (1 sentence: who it's for + key benefit) - Microcopy under the CTA button (1 line, defuses the biggest objection, e.g. "Free trial, no credit card") The 5 must use 5 distinct angles: outcome / contrast / counter-conventional-wisdom / tool / scenario. Do not start more than one with the same verb.Selection rule: read each headline aloud. If your tongue trips, drop it. Then check it fits the 2026 hero bar — under 8 words, no jargon a stranger would have to decode.
-
Launch email (with subject A/B):
Based on the brief, write a launch email: - Subject: 3 variants (benefit / curiosity / identity), 30-50 characters each; the key word in the first 40 characters - Pre-header: 1 sentence <=80 chars, not repeating the subject, completes the subject's thought - Opening line: standalone hook - Body: 200 words, structure = pain 1 graf + product enters 1 graf + 1 concrete use case + CTA - Primary CTA + secondary CTA (for readers who won't buy now but might browse) - Mobile format: <=2 lines per paragraph, links on their own lineWhy those subject limits: over 60% of opens are on mobile, and Gmail’s app truncates around 30 characters while an iPhone in portrait shows roughly 41. Anything past ~40 characters is a gamble.
-
Press blurb:
Based on the brief, write 2 versions for press: - 50-word version (for journalists who skim): what + who-for + when + differentiator - 350-word version: opening news graf (lead = 25-35 words, answers the 5 Ws) + 1 founder quote (conversational, no marketing speak) + launch info (date / price / how to trial) + media contact placeholder Keep the long version under 400 words total. Banned: "is excited to announce" / "today we unveiled".The 400-word ceiling is not arbitrary: most journalists give a release 5–10 seconds and prefer it under 400 words. Put the news in the first sentence.
-
Demo script (60 seconds):
Based on the brief, write a 60-second product demo voiceover: - Total length <=150 words (about 60 seconds of natural narration) - 5 cue points, ~12 seconds each, format: [00:00-00:12] cue: [on-screen action] vo: [exact line] - 0-12s must show the problem first, then the product enters - 50-60s leaves a CTA (signup URL / how to try) -
FAQ (8 questions):
Based on the brief, write 8 launch FAQ items: - First 3: objection-type (price / learning curve / vs competitor X) - Middle 3: usage-type (can-I / how-do-I / what's the limit) - Last 2: trust-type (data / privacy / how to cancel) - Each answer <=2 sentences: lead with "Yes / No / We do", then 1 sentence of reasoning - Don't invent features that don't exist; don't dodge — name limits explicitly -
Human pass (don’t skip). Do exactly these four things:
- Rewrite the opening sentence of every piece. AI openings are reliably generic; write your own.
- Add one detail to each piece that only you would know: a real user name, a concrete number (not “many”), an inside reference.
- Style: straight quotes
"/'to curly"/'; em-dash--to—; trim Oxford commas to your house style. - Readability pass: run each piece through Hemingway App and cut every sentence above Grade 9. Kill adverb stacks (“really”, “very”, “honestly”).
-
If this is a version drop rather than a brand-new product, run the same brief through the AI release notes drafter for the in-app announcement page.
Recommended workflow
Brief the AI once, generate all six pieces in the same thread, run the read-aloud check, write your own openings and add one insider detail per piece, verify mobile formatting, then schedule. If part of the funnel runs through Xiaohongshu, write the thumbnail-legible cover lines with our AI Xiaohongshu cover-text workflow — explore-feed click-through depends on those 8-character cover lines more than on the post title.
Which model for which piece
All three frontier models clear this work, and the brief matters far more than the engine. As of June 2026, the practical split:
- Claude (Opus 4.7 / Sonnet 4.6) — best for the announcement, email, and FAQ, where natural tone and low-edit first drafts save the most time. Both models carry a 1M-token context, so they hold a long brief without drifting.
- ChatGPT (GPT-5.5) — fastest at throwing off five hero angles or three subject variants when you want raw quantity to choose from.
- Gemini 3.1 Pro — useful if you are pasting in a long existing doc (old landing page, support transcripts) and want the copy grounded in it; 1M-token context handles the dump.
Pick one, keep the whole launch in a single thread, and let the human pass — not a model swap — do the polishing.
Common mistakes
- Different story spine per piece. Six assets that each tell a slightly different story confuse the reader. One brief fixes this.
- Shipping the AI’s opening line. It is the single most “AI-flavored” sentence in any draft. Rewrite all six.
- Skipping the FAQ. The questions arrive anyway, just in your inbox and replies instead of on the page.
- No demo script. A video launch without a 150-word script becomes a two-minute ramble nobody finishes.
- Letting the press blurb run long. Past 400 words, journalists stop reading.
FAQ
- How early should I start? Five days before launch is comfortable, three days is the minimum, two days is survivable but rushed.
- Should I A/B test launch copy? Test email subject lines (they are cheap to vary and 30–50 characters is the proven band); for everything else, commit and learn for the next launch.
- Why does the email subject have to be 30–50 characters? Over 60% of email opens are on mobile, and Gmail’s app shows about 30 characters while an iPhone in portrait shows roughly 41. Put the key word in the first 40 characters or it gets truncated.
- Why cap the press release near 400 words? Surveys of journalists in 2026 show most prefer releases under 400 words and spend 5–10 seconds deciding whether to read on. Length signals you don’t understand their job.
- Which AI should I use? Any of the three works. Claude needs the least rewriting on prose, ChatGPT is fastest for generating variants, and Gemini is handy when you want the copy grounded in a long doc you paste in.
- Can I reuse this brief later? Yes. Keep the filled-in brief from step 1; for a version drop, feed it straight into the release notes drafter instead of starting over.