TL;DR
- Use AI to structure the page (hook → pain → included → proof → FAQ → pricing → CTA) and to vary angles, not to calibrate tone — that part you control with explicit instructions.
- Each format has a different load-bearing section: a course lives on its module table, a template pack on its file manifest + previews, an ebook on its full table of contents + free sample chapter. One prompt across all three is the most common mistake.
- Tell the model the audience temperature (“warm list, soft language, no urgency theatre”) and a banned-phrase list. Warm audiences read “transform your life” and “only 3 spots left” as a trust signal in reverse.
- Self-serve product landing pages convert at a 4-10% median (best-in-class 12-18%), per Unbounce/landing-page benchmark data (2026). A clear, honest page beats a high-pressure one on a list you’ve earned.
- Where you host matters for take-home revenue: as of June 2026, Gumroad charges 10% + $0.50 per direct sale, Lemon Squeezy 5% + $0.50 (and is your merchant of record for tax), Payhip 5% free / 0% on its $99/mo plan (you handle most tax). Pick before you write the pricing block.
The task
You are launching a $30-300 digital product — a course, template pack, ebook, or system. You want a one-page sales page that converts your warm audience (your list, your followers) without sounding like an internet marketer’s funnel. The risk is that AI defaults to aggressive copy (“transform your life,” “limited spots”) that burns trust with the exact people most likely to buy.
This is a structuring job, not a “press generate” job. AI drafts a clean skeleton in seconds; you supply the proof, the price, and the voice.
Pick where you’ll sell before you write the page
The pricing block and the guarantee copy depend on your platform’s fees and refund rules, so decide this first. Current rates as of June 2026:
| Platform | Per-sale fee | Monthly | Merchant of record (handles tax) | Best when |
|---|---|---|---|---|
| Gumroad | 10% + $0.50 direct (30% via Discover) | $0 | Yes (since Jan 2025) | Simplest start; you accept the 10% |
| Lemon Squeezy | 5% + $0.50 (+1.5% intl/PayPal) | $0 | Yes | You want MoR tax handling at half Gumroad’s rate |
| Payhip | 5% free / 2% at $29 / 0% at $99 | $0-99 | No (EU/UK VAT only) | You’re past ~$2,000/mo and tax compliance is manageable |
Sources: Gumroad pricing, Lemon Squeezy 2026 update. Stripe card processing (≈2.9% + $0.30) stacks on top for platforms that aren’t MoR. The practical takeaway: on a $99 product, Gumroad keeps ≈$10.40 of fees, Lemon Squeezy ≈$5.45. Write a price the model can quote, then a guarantee that matches the platform (Gumroad does not refund its fee on a refunded sale, so don’t promise “no-questions-asked, full money back” if you can’t absorb that).
Each format demands a different page
Courses, template packs, and ebooks share the “hook → pain → included → proof → FAQ → pricing → CTA” skeleton, but the load-bearing section differs in each. Reusing one prompt across all three formats is the single most common mistake.
Course landing
- Hero: outcome-led headline. “Ship your first SaaS in 30 days,” “Finish your book in 4 weeks.” Sell the result, not the module count.
- Curriculum table is the sales tool: the module table (M1-M6 with per-module outcome and deliverable) goes mid-page. Buyers read it end to end.
- Instructor authority: track record, past student data, alumni outcomes (“237 students shipped a product”). At course prices, your credibility is the product.
- Cohort vs evergreen: cohort pages lead with “Next cohort June 3” + peer group; evergreen leads with “Start tonight” + async support model. These are not the same page with one swapped line.
- Social proof: testimonials anchored to specific outcomes (“Week 4 method got me my first $499 sale”), not vague praise.
Copy-ready prompt:
Draft a course sales page.
Course: [name]
Price: [$ amount]
Modules + total runtime: [6 modules / ~4 hours]
Instructor track record: [one line]
Past alumni outcomes: [specific numbers + named students]
Cohort or evergreen: [tag]
Audience: [warm / cold]
Structure:
- Hero: outcome-led headline + sub-line
- "What you'll have at the end" (3 concrete nouns, not adjectives)
- Module table (M1-Mx: outcome + 3 lessons + deliverable)
- Instructor authority section
- Alumni outcome testimonials (named + specific numbers)
- Cohort vs evergreen block (start date OR instant access)
- FAQ + pricing + guarantee + CTA
Constraints: warm-audience tone; no "transform your life"; module
table appears above the fold of the mid-page.Template-pack landing
- Hero: preview grid. A template is a visual product; six thumbnails do more work than a thousand words.
- “What you get” file manifest: exact file counts. “57 .figma source files + 12 PSDs + 1 Notion dashboard + 1 README.” Vague “complete template kit” copy makes buyers bounce.
- Use-case mapping: which template fits which scenario. “Sales page → Templates A / B / C; landing page → D / E; pricing page → F.” Buyers buy when they can mentally drop a template into their own project.
- Tool compatibility: Figma version, whether the Notion free tier works, PSD vs
.ai. One line, prominently. A large share of refunds otherwise come from “won’t open.” - Previews, previews, previews: at least 8 high-res screenshots with hover-to-zoom.
Copy-ready prompt:
Draft a template-pack sales page.
Template pack: [name]
Price: [$ amount]
File manifest (exact counts): [57 figma + 12 PSD + 1 Notion]
Tool/version requirement: [Figma v.x / Notion free works]
Preview count: [x screenshots]
Typical use cases: [3-5 scenarios]
Structure:
- Hero: 3x2 preview grid + one-line promise
- "What you get" — file manifest (exact counts + file types)
- Use-case mapping table (scenario -> which template)
- Tool compatibility line
- Large preview gallery (hover to zoom)
- Testimonials anchored to "I used Template X to ship Y"
- FAQ + pricing + CTA
Constraints: every claim must be provable from a preview; no
"professional grade," "premium," and other empty adjectives.Ebook landing
- Hero: promise + author authority. “60,000 words and 3 years of SaaS pricing in the field.” The book needs weight.
- TOC is the sales page: the chapter list with one-line abstracts per chapter sits mid-page. Buyers decide from the TOC.
- Free sample chapter: one PDF link, 10-15 pages. “Read the sample before buying” outconverts a teaser modal.
- Format notes: PDF / ePub / Kindle, all three. One line: “All three formats delivered by email within 5 minutes of purchase.”
- Author authority: not just a CV. Include “why you wrote this”: 3 years in the trenches, 27 founders interviewed, original data.
Copy-ready prompt:
Draft an ebook sales page.
Ebook: [title]
Price: [$ amount]
Chapter count + total word count: [12 chapters / 60k words]
Author authority: [3 years SaaS + 27 founder interviews]
Formats: [PDF + ePub + Kindle]
Sample chapter PDF URL: [URL]
Structure:
- Hero: promise headline + sub-line (the book's weight)
- Author authority section ("why I wrote this")
- Full TOC (every chapter with one-line abstract)
- Sample chapter download CTA (PDF link)
- Testimonials with specific reader takeaways
- Format notes (PDF / ePub / Kindle)
- FAQ + pricing + CTA
Constraints: the full TOC must appear on the page, not "buy to see
full TOC"; sample chapter CTA goes above the fold.ChatGPT or Claude for this?
Either ships a usable draft. The practical difference for sales copy, as of June 2026:
- Claude (Sonnet 4.6, free tier or $20 Pro): tends to default to calmer, less hype-y prose, which is exactly what a warm audience wants. Good first choice when your worry is over-selling. Opus 4.7 ($100 Max) is worth it only for long, multi-product launches.
- ChatGPT (GPT-5.5, $20 Plus): strong at generating angle variations fast — give it your draft and ask for “five alternate hero headlines, each a different emotional register.” Useful for the A/B testing step.
For a single $30-300 page, a free or $20 tier handles it. Don’t pay for Pro/Max tiers just to write one landing page. Either way, paste a real example of a sales page in your voice as a style anchor; both models match a sample far better than they follow an adjective list.
When AI helps — and when it does not
AI is good at structuring sales pages and varying angles per section. It is poor at calibrating tone for your audience. Warm audiences want soft, value-first language; cold traffic tolerates more pressure. State it explicitly: “warm audience, soft language, no urgency theatre.”
What to feed the AI
- Product description and price
- Target buyer (warm vs cold) and their pain, in their own words
- What’s included (concrete deliverables: modules, files, pages)
- Proof: testimonials, early-user numbers, your relevant credentials
- Guarantee policy (matched to your platform’s refund rules)
- Brand voice and banned phrases (“transform,” “10x,” “limited spots”)
Copy-ready prompt (any format)
Draft a sales page.
Product: [one line]
Price: [amount]
Audience: [warm / cold]
Audience pain: [one line]
Included (concrete): [list]
Proof: [testimonials, numbers, credentials]
Guarantee: [14-day / 30-day / none]
Brand voice and banned phrases: [list]
Structure:
- Hook headline + sub-line
- Pain-name section (3 paragraphs - name the pain in your
audience's words)
- What's included (visual bullets - modules, files, pages, time saved)
- Proof block (testimonials with attribution, numbers)
- FAQ (5 questions handling real objections)
- Pricing + guarantee (clear, no hidden details)
- Final CTA (button copy + risk-reducer)
Constraints:
- Warm audience - soft language, no urgency theatre
- Every claim has evidence or it doesn't appear
- No "transform your life," "10x," "limited spots"
- Cover real objections in FAQ, not generic onesFor higher-ticket ($300-1,000): “Add a ‘who this is for / who this is NOT for’ section near the top.”
Recommended output structure
Hook → pain-name → included → proof → FAQ → pricing → CTA. Pricing visible and unhidden. Guarantee near the CTA. The “who is not for this” disclaimer adds trust at higher price points.
How to check the output is usable
- Every claim has evidence or a placeholder marked
[VERIFY] - “Included” is concrete (modules, file counts, page counts)
- Proof block uses real testimonials with names (or marked
[NEED]) - FAQ handles your real objections, not generic ones
- No urgency theatre. Warm audiences detect it
- Pricing is visible without scrolling past 3 sections
- Guarantee copy matches your platform’s actual refund policy
Common mistakes
- Marketing-speak (“transform your life”) that burns trust
- Hiding what’s included. Buyers detect padding
- No FAQ, which leaves real objections unanswered
- Fake urgency. Burns trust with warm audiences forever
- A long page for a simple product. Match length to ticket
- Letting AI invent testimonials. Verify or remove
- Promising a guarantee your platform’s refund rules can’t support
FAQ
One page or a multi-page funnel? One page for warm audiences who already know you. Funnels earn their complexity on cold traffic, where you have to build trust across steps.
Should the price be hidden? No. Warm audiences read “request pricing” or “DM for price” as a sign you’re about to negotiate them up. Show the number.
What about a downsell? Useful for cold traffic; less needed for warm. A warm list converts on the main offer or not at all — a downsell rarely rescues it.
What conversion rate should I expect? For a warm list, a clear self-serve product page commonly lands in the 4-10% range, with strong pages reaching 12-18% (landing-page benchmarks, 2026). Cold traffic is far lower; judge against your own baseline, not a screenshot from a course seller.
Can I let AI write my testimonials? No. Invented testimonials are a refund and reputation risk, and platforms treat them as policy violations. Use real quotes with names, or mark the slot [NEED] and fill it before launch.
Related
- Sales copy prompts — alternative sales-copy phrasings
- Landing page copy prompts — landing page variants
- Landing page copy — general landing page workflow
- Pricing page copy — tiered pricing
- Newsletter lead magnet AI — lead magnet to drive list growth
- Email marketing copy — promo emails driving to the page
- Brand story prompts — the story that anchors the pain section
Tags: #AI writing #Content creation #Creator #Digital product #Sales copy