Landing page prompts for ChatGPT and Claude

These 14 landing page prompts take you from a blank page to a tested one: plan the structure, write the hero, answer objections, build the FAQ, draft A/B variants and check every claim before launch. They work in ChatGPT, Claude and Gemini. Each prompt asks for your real product facts and customer language, because a page built from a one-line product description sounds like every other page.

How to use these prompts

Copy a prompt, replace the blanks in double braces, or keep the default after the colon: {{goal:start a free trial}} means "start a free trial" unless you change it. Blanks with no default, like {{facts}}, need your own material.

Before you start, collect:

• One goal: the single action the page asks for. Two goals usually means two pages.
• Facts: features, prices, limits, guarantees and any numbers you can prove.
• Customer language: phrases from reviews, support tickets, sales calls and interviews.
• Objections: the reasons people give for not buying, in their words.

Run the prompts in order the first time. The structure prompts decide what the page needs to say; the copy prompts then fill it in. For sharper headlines and offers beyond the landing page, use the copywriting prompts. For button labels, form errors and empty states, use the UX writing prompts.

Ask for options, not one answer. It is far easier to pick and combine from eight headlines than to fix one.

Which prompts plan the landing page structure?

1. Page brief from one goal

Writing the brief first stops the page from trying to do everything.

Help me write a one-page brief for a landing page.

Product: {{product:a time-tracking app for freelancers}}
Goal of the page: {{goal:start a free trial}}
Traffic source: {{traffic:Google search ads}}
Audience: {{audience:freelance designers}}

Facts I can prove:
{{facts}}

Give me: the visitor's main problem, what they already believe when they
arrive from this traffic source, the one promise the page must make, the
proof that backs it, the top 3 objections, and what "success" means for
this page. Mark anything you inferred rather than took from my facts.

2. Section outline by awareness level

A visitor who has never heard of you needs a different page from one comparing you to a competitor. This makes the order fit.

Outline the sections for a landing page for {{product:a time-tracking app for freelancers}}. The visitor is {{awareness:aware of the problem but not of our product}}.

For each section give: its job in one sentence, what it must contain, and
which fact or quote from my notes it should use. Put the sections in the
order this visitor needs them. Say which sections you would drop for a
visitor who already knows the product.

Notes:
{{facts}}

3. Five-second test of the top of the page

Paste your current hero and the first section. If a stranger can't say what it is and who it's for, nothing below matters.

Here is the top of my landing page as a visitor sees it before scrolling.

{{hero_text}}

Pretend you are {{audience:a freelance designer}} who just clicked an ad.
Answer in one line each: What is this? Who is it for? What should I do next?
Why should I believe it? Then list every word or phrase that a first-time
visitor might not understand, and suggest plainer wording.

Which prompts write the hero section?

4. Hero headline and subhead options

The hero carries most of the page. Different angles give you real choices instead of eight rewordings of the same line.

Write 10 hero headline and subhead pairs for {{product:a time-tracking app for freelancers}}, aimed at {{audience:freelance designers}}.

Use 5 different angles, two pairs each: the outcome, the pain removed, the
comparison with {{alternative:a spreadsheet}}, a specific number from my
facts, and the customer's own words. Headlines under 10 words, subheads
under 25. Label each pair with its angle. Use only facts below.

Facts:
{{facts}}

Customer phrases:
{{customer_quotes}}

5. Call to action button and the line under it

The button says what happens next. The microcopy under it removes the last worry.

Write 8 call to action options for a button that {{action:starts a 14-day trial}}.

Each option: button text (2 to 5 words, starting with a verb, saying what
the visitor gets) plus one short line under the button that answers the
most likely worry, such as cost, commitment or effort. Use only these
facts about the offer: {{offer_facts}}. Don't promise anything not listed.

6. Rewrite the hero from customer quotes

When the hero sounds like marketing, rebuild it from how customers describe the problem.

Here are quotes from customer interviews and reviews:
{{customer_quotes}}

1. List the 5 phrases customers use most to describe their problem or the
   result they wanted, with how often each appears.
2. Write 5 headlines that reuse those phrases as closely as possible.
3. Tell me which current headline words never appear in what customers say.

Current headline: {{current_headline}}

How do you handle objections on a landing page?

7. Find the real objections

Objections you guess are usually the wrong ones. This pulls them from what prospects actually said.

Below are notes from sales calls, support tickets and cancelled-trial
surveys for {{product:a time-tracking app for freelancers}}.

List every reason someone hesitated or didn't buy. Group them into themes,
count how often each theme appears, and quote one example per theme word
for word. Rank by frequency. Do not add objections that aren't in the
notes.

Notes:
{{notes}}

8. Write the objection section

Answer the top objections directly, with proof, near the point on the page where they come up.

Write a landing page section that answers these objections for
{{audience:freelance designers}}:

{{objections}}

For each: a short heading in the visitor's words, a 2 to 3 sentence answer,
and the proof from my facts (a number, a policy, a real quote). If I have no
proof for an answer, write "PROOF NEEDED" instead of inventing it. Suggest
where on the page each answer should appear.

Facts:
{{facts}}

9. Place real social proof

Use only quotes you have permission to publish. This prompt chooses and trims; it never writes testimonials.

Here are real customer testimonials I have permission to use:
{{testimonials}}

Pick the best quote for each of these spots: under the hero, next to
pricing, next to the objection "{{objection:it takes too long to set up}}",
and above the final call to action. You may shorten a quote but not change
its meaning; show the original next to the shortened version. Flag any
quote that describes results most customers wouldn't get.

Which prompts write a landing page FAQ?

10. FAQ from support tickets

The best FAQ answers what people actually ask before buying. Support tickets and pre-sales emails are the source.

From the support tickets and pre-sales emails below, write a landing page
FAQ with {{count:8}} questions.

Use the customer's wording for each question. Answers: 2 to 3 sentences,
direct, starting with the answer. Use only facts from my notes; if a
question needs a fact I didn't give, list it under "Facts needed" instead
of answering. Order the questions by how often they come up.

Tickets and emails:
{{tickets}}

Facts:
{{facts}}

11. Pricing and refund FAQ

Money questions need exact answers. This prompt checks the answers against your actual terms.

Write FAQ answers about pricing, billing, cancellation and refunds for
{{product:a time-tracking app for freelancers}}.

Our actual terms, which are the only source you may use:
{{terms}}

For each answer, quote the part of the terms it relies on. If the terms
don't cover a common question (for example proration or taxes), list it as
a gap for me to decide. Plain language, no legal jargon.

How do you write landing page A/B test variants?

If your test uses separate URLs, follow Google's guidance on website testing: don't show Googlebot different content from people, point variants to the original with rel="canonical", use a 302 (temporary) redirect rather than a 301, and run the experiment only as long as necessary.

12. Hypothesis and variants

A test without a hypothesis teaches you nothing even when it wins.

I want to A/B test the {{element:hero headline}} on my landing page.

Current version: {{current}}
Page goal: {{goal:start a free trial}}
What I know about visitors: {{insight}}

Write 3 hypotheses in the form "Because [evidence], changing [element] to
[change] will increase [metric] for [audience]." For each, write the
variant. Make the variants differ in their promise or angle, not just
wording. Tell me which one to test first and why.

13. Read the results

Paste the numbers from your testing tool. The point is to know what the result does and doesn't prove.

Here are the results of my A/B test:

Variant A: {{a_visitors}} visitors, {{a_conversions}} conversions
Variant B: {{b_visitors}} visitors, {{b_conversions}} conversions
Test ran for: {{duration:14 days}}
My tool reports: {{tool_result:the significance or confidence figure}}

Explain in plain language what this result means, whether the sample is
large enough to trust, and what could make it misleading (novelty, seasonal
traffic, uneven traffic sources). Suggest the next test either way.

How do you check a landing page before it goes live?

The US Federal Trade Commission's advertising guide for small businesses says advertisers need objective evidence for their claims before the ad runs, and that this covers implied claims as well as the words on the page. It also says that endorsements from consumers must reflect the typical experience, and that adding "results may vary" is not enough. The FTC's Consumer Reviews and Testimonials Rule went into effect on October 21, 2024 and covers fake or false reviews and testimonials. This prompt is a practical check, not legal advice.

14. Claim check

Run it on the final draft. It finds the claims you forgot you were making.

Review this landing page copy before launch.

1. List every claim it makes, stated or implied (speed, savings, results,
   comparisons, "best", "only", numbers, guarantees).
2. For each claim, point to the fact in my evidence that supports it, or
   mark it "NO EVIDENCE".
3. Flag any testimonial that shows results a typical customer wouldn't get.
4. Suggest a supported rewrite for every claim marked "NO EVIDENCE".

Copy:
{{copy}}

Evidence:
{{evidence}}

Which landing page tasks should stay human?

How do you reuse these prompts for every new page?

Most teams build landing pages again and again: one per campaign, segment or feature. Kipt keeps prompts like these as Markdown in a private GitHub repo you own, with every blank ready to fill in, tags for finding them, and Cmd+K search. The claim check is a good one to turn into an Agent Skill with the free SKILL.md generator, so it runs every time an agent drafts a page. For launch emails and announcements, see the product launch prompts.