A landing page dies one block at a time: a hero that explains what the product is instead of what it does, a social-proof strip with no numbers and no names, a feature grid that reads like a competitor’s, and two CTAs at the bottom fighting each other. The median landing page converts about 4.3% in 2026 (per Unbounce’s analysis of 41,000 pages), and the top 10% convert above 11% — roughly three times the average. The gap between those two numbers is mostly block-by-block discipline. These 12 prompts force each block to earn its space — hero earns the scroll, social proof earns trust, problem-agitate earns four nods, features earn comprehension, and the final CTA earns the click.
TL;DR
- Run these 12 prompts in order; each one owns a single block and a single job, so the page reads as one argument instead of a pile of sections.
- Feed real evidence (named customers, quantified outcomes, exact pricing) — the prompts are scaffolding, not a substitute for facts. A named-customer claim lifts conversions ~22% vs ~8% for a bare logo strip.
- Above the fold is non-negotiable: Nielsen Norman Group found content above the fold gets 84% more attention, and the effective “is this for me?” window is ~5 seconds in 2026.
- Best model for this work as of June 2026: Claude Opus 4.7 or Sonnet 4.6 for voice-consistent long-form copy; GPT-5.5 Thinking for structured comparison/pricing logic. Either handles all 12 prompts.
Best for
- Product launches
- B2B landing pages
- DTC product pages
- Indie SaaS pricing pages
- Newsletter / community landing pages
Which model to run these in
All 12 prompts are model-agnostic, but copy quality varies. As of June 2026:
- Claude Sonnet 4.6 (free tier, or Claude Pro $20/mo) — the workhorse for brand-voice copy. Paste your existing site copy in the same chat first so it matches tone. Best default for the hero, problem-agitate, and FAQ blocks.
- Claude Opus 4.7 (Claude Max, $100/mo) — when you need the sharpest hero headline or the most ruthless feature edit; it cuts filler harder than Sonnet.
- GPT-5.5 Thinking (ChatGPT Plus $20/mo, picker set to Thinking) — strongest on the comparison-table and pricing logic, where you want it to reason about which dimensions actually matter.
- Gemini 3.1 Pro (Google AI Pro $19.99/mo) — useful if your evidence lives in Google Docs/Sheets you can attach directly.
Whatever you pick, keep one chat per page so the model carries voice and claims across all 12 blocks.
1. Hero section
Write the hero section for "{product}". Audience: {audience}. Output: H1 (≤8 words), sub-headline (≤20 words), 1-line clarifier on what it IS (1 noun), and CTA button copy + secondary CTA. Avoid jargon.The H1 answers one question — “Is this for me, and does it solve my problem?” — and it has to do it in roughly 5 seconds. If your draft H1 describes a category instead of an outcome, send it back.
2. Social-proof section
Write the social-proof section. Available evidence: {logos, numbers, quotes}. Output: a headline ("Trusted by X who Y"), a quote with attribution, 1 quantified outcome, and a logo strip caption. Avoid "loved by everyone".Specificity is the multiplier here. A named-customer claim with context (“Used by 8 of the Fortune 50”) lifts conversions roughly 22%, versus about 14% for a single testimonial card and 8% for an unlabeled logo strip. Three to five testimonials is the sweet spot before cognitive load eats the gain.
3. Problem-agitate section
Write the problem section. Audience's current pain: {paste}. Output: a 1-sentence headline that mirrors the pain, 3 specific pain bullets, 1 cost-of-inaction line. Should make the reader nod 4 times.4. Solution-overview section
Write the "how it works" overview after problem. Output: 3 steps (verb + object), each 1 line, with the 1-line outcome under each step. Total ≤60 words. End with a transition to features.5. Feature-block section (3 features)
For 3 features of {product}, write feature blocks. Each block: headline (≤6 words, value-led not feature-led), 1-sentence explanation, 1 supporting fact / metric, 1 visual placeholder description.6. Use-case row (for multi-audience products)
{Product} serves {3 audiences}. Write a use-case row for each: audience name, their job, how {product} helps in 1 sentence, link text to detail page. Keep parallel structure.7. Comparison-table section
Write a comparison-table section for {product} vs {top 2 competitors}. 5 rows of meaningful dimensions (not "advanced features ✓"). For each, the honest answer per column. End with a 1-line "pick us if you need X".This is the block GPT-5.5 Thinking handles best — ask it to reason about which 5 dimensions a buyer actually weighs, then fill the table. Honest “no” rows build more trust than an all-green column.
8. Pricing section
Write the pricing section for {N} tiers. For each: tier name, who it is for, 3 included items, 1 not-included item, CTA. End with a "questions about pricing?" link. No fake scarcity.The “who it is for” line is the highest-leverage edit on the whole page. Without it, buyers stall on the wrong tier and bounce. Drop fake countdown timers — they erode trust faster than they create urgency.
9. FAQ section
Write the FAQ section for {product}. Mine likely buyer hesitations: {paste}. Output 8 Q&A pairs, each answer ≤60 words. Include 1 question that handles the strongest objection.10. Risk-reversal section
Write the risk-reversal section above the final CTA. Include: trial / guarantee terms, cancel policy, support response time, 1 specific reassurance. ≤70 words total.11. Final-CTA section
Write the final CTA section. Output: a 1-sentence restatement of the value, 1 secondary trust signal, the button copy (different from hero), and 1 alternate path ("not ready? read the case study"). ≤50 words total.12. Section-order audit
Below is my current landing page section order. Evaluate whether the order matches buyer-decision flow (awareness → consideration → commitment) for my audience: {audience}. Propose a new order with rationale.
{paste current sections}Common mistakes
- Hero that explains what the product IS instead of what it DOES for the reader
- Generic feature blocks that read like a competitor’s marketing site
- Social proof without a number, a name, or anything falsifiable
- Pricing tiers without “who this is for” — buyers stall on the wrong tier
- Two CTAs competing at the bottom, splitting attention and ad-attribution
- FAQ that dodges the strongest objection instead of answering it directly
- Section order that doesn’t follow buyer-decision flow (problem → proof → solution → ask)
- A heavy hero image that tanks LCP; compress to WebP so the page is usable inside the 5-second window
FAQ
Which AI model writes the best landing-page copy in 2026? For brand-voice copy, Claude Sonnet 4.6 (free or Claude Pro $20/mo) is the best default; paste your existing copy first so it matches tone. For the comparison-table and pricing logic, GPT-5.5 Thinking (ChatGPT Plus $20/mo) reasons more cleanly about which dimensions matter. Both clear all 12 prompts.
In what order should I run these prompts? Follow the numbering — it mirrors buyer-decision flow: hero → social proof → problem → solution → features → use cases → comparison → pricing → FAQ → risk-reversal → final CTA. Run prompt 12 last to audit the whole sequence against your specific audience.
What conversion rate should a new landing page target? The 2026 median is about 4.3% across industries; ecommerce sits near 3–4%, SaaS free-trial pages near 7%, and high-intent lead-gen near 8%. Top-10% pages clear 11%. Treat the median as a floor, not a goal, and remember traffic source moves the number more than industry — pages built for a specific email campaign convert far better than generic paid-search landers.
Will AI-written copy hurt my SEO or AdSense approval? Only if you ship it raw. The risk is generic, unverified claims. These prompts force specifics — named customers, real metrics, exact pricing — which is exactly what reviewers and search engines reward. Always fact-check every number the model produces against your own data before publishing.
How many testimonials should the social-proof block show? Three to five on the page itself strikes the best balance between credibility and cognitive load. Keep a deeper library elsewhere, and prefer named customers with revenue or outcome context over anonymous five-star blurbs.
Related
Tags: #Prompt #E-commerce #E-commerce