Products SEO & GEO Agent Paid Ads Agent Website Agent Conversion Agent
How it works Pricing FAQ Blog Book a call

ChatGPT for landing pages: prompts to improve copy and design

Updated · 8 min read

Want help applying this to your business? Talk it through on a free 30-minute call.

Book a call

ChatGPT for landing pages: what it can and can't do

ChatGPT reads a page the way a stranger would and writes ten versions of a line in seconds. It knows nothing about your business, prices or visitors unless you tell it. OpenAI's prompting guide says its GPT models "benefit from precise instructions that explicitly provide the logic and data required to complete the task in the prompt."

JobCan ChatGPT do it?What it needs from you
Spot what's unclear or missingYesA phone screenshot and the page text
Rewrite headline and buttonYesOffer, price, area, real customer words
Make the page match an adYesThe ad text and its keyword
Write an FAQ sectionYes, from your facts onlyThe questions customers ask you
Suggest what to testYesMonthly visits and leads
Write reviews or testimonialsNo: invented ones break the lawReal reviews, quoted as written
Measure speed, watch visitorsNoPageSpeed Insights, analytics, heatmaps
Pick the winning versionNoLeads from each version

First write one page of facts: what you sell, where, the price or range, guarantees and license numbers (only what's true), the one action you want, the ad that sends people to the page, 5 to 10 real reviews, and the questions people ask before booking. Paste it at the top of each chat, or keep it in a ChatGPT Project, which OpenAI's pricing page lists on every individual plan, Free included.

Prompt 1: review the page as a customer would

Find problems before writing anything new. Screenshot the page's first screen on your phone and paste the full text below it. OpenAI's docs warn that "Vision models can make mistakes" and advise you to "Enlarge text within the image to improve readability," so send one screen per image.

Prompt: "You are a [type of customer] in [city]. You just tapped an ad that said '[ad headline]'. Attached is the first screen of the page on a phone; the full text is below. In three lines: what does this business offer, where, and what should you do next? Then list the 5 things most likely to make you hesitate or leave, worst first, quoting the words or saying where they are. Don't rewrite anything yet."

Layout follow-up: "From the screenshot only: what does your eye land on first, and is the button visible without scrolling? Suggest the order of sections, top to bottom, for someone who clicked this ad."

Form follow-up: "Now go through [booking / the quote form] as a first-time visitor on a phone. List each field and tap, and mark where you'd give up."

If the three-line summary doesn't match what you sell, no button color will fix the page.

Prompt 2: write the headline from your customers' words

The best headline material is already in your reviews: the result customers mention, in their words. Ask ChatGPT to find it, not invent it.

Prompt: "Below are my fact sheet, 10 customer reviews and the questions people ask before booking. Find the 3 results customers mention most, in their own words. Write 8 headlines of 10 words or fewer, each naming the service and the place or the result, with a one-line subheading. Use only facts from the fact sheet and reviews."

Shortlist two and keep the one you'd say to a customer on the phone.

Prompt 3: make the page match the ad

People who click an ad expect the page to keep its promise. Google tells advertisers:

"Choose a landing page that closely matches your ad and keywords." (Google Ads Help: Optimize your ads and landing pages)

The page should also "mirror the call-to-action in your ad text," Google adds.

Prompt: "Here is my Google ad: [headlines and descriptions]. Keyword: [keyword]. Here is the first screen of the page it opens: [text]. List each promise the ad makes and whether the first screen keeps it. Then rewrite the headline, subheading and button so someone who clicked this ad sees the same offer at once."

Run it for each ad that opens a different page. More: Landing page for Google Ads.

Prompt 4: buttons and forms

A call to action (CTA) is the button that tells visitors what to do next. "Submit" says nothing; "Get my free estimate" says what happens after the tap.

Prompt: "The action I want is [book a free estimate / call now / pick a time]. Write 10 button labels of 5 words or fewer that say what happens next. Use 'free' or 'no obligation' only if my fact sheet says so. Then mark each of my form fields [list] as 'needed to reply' or 'can wait for the call'."

Prompt 5: answer objections without inventing facts

Visitors leave with unanswered questions about price, area or timing. An FAQ block fixes that, but this is where ChatGPT most often makes things up. Google explains why:

"Keep in mind that generative models don't retrieve facts, but predict a likely sequence of words based on their training data." (Google Search Central: Guidance on using generative AI content)

Google calls it "critical to manually factcheck and review all AI-generated content" before publishing, page title and meta description included.

Prompt: "Here are questions customers ask before booking: [list]. Answer each in 2 sentences or fewer, using only my fact sheet. If the fact sheet doesn't answer it, write NEEDS ANSWER instead of guessing."

Each NEEDS ANSWER is a gap your page had all along.

Prompt 6: plan a test and read the results

ChatGPT can suggest what to test; only your visitors can say what won. An A/B test splits visitors between two versions and compares leads. With a few leads a month, a real difference can take months to show: when a test is worth running.

Prompt: "My page gets about [number] visits and [number] leads a month. Suggest 5 changes to test, biggest likely effect first: what changes, why, and what number would show it worked. Then tell me plainly whether my traffic can show a difference in 6 to 8 weeks, and if not, which changes to simply make."

To show it what visitors did, upload an export, such as a Google Analytics report saved as a CSV file, and ask what the numbers show and can't show. OpenAI's pricing page lists file uploads and data analysis on Free, with limits. For where people tap and stop, use Microsoft Clarity's heatmaps; Microsoft says Clarity "is a free service forever." Delete names, emails and phone numbers from any export first.

What ChatGPT must not write on your landing page

Reviews and testimonials. In the US, the FTC's 2024 rule bans a business from writing or creating a review or testimonial that misrepresents that the reviewer exists or used the service (16 CFR 465.2). The FTC's announcement names "AI-generated fake reviews" as an example. In Canada, publishing a testimonial to promote a business is reviewable conduct under the Competition Act unless the person published it first or approved it in writing, and your version matches what they said (section 74.02). Quote real reviews word for word and say where they came from.

Facts it can't know. Prices, years in business, warranties, "#1 in town", statistics. Check each against your fact sheet.

Fake urgency. "Only 2 spots left" must be true. The US and Canadian rules: Landing page best practices.

Speed verdicts. ChatGPT doesn't measure load time. Google's PageSpeed Insights does, in a simulated test, plus real-visitor data when your site gets enough Chrome traffic. Fixes: Core Web Vitals.

Customer data. OpenAI's pricing page lists "Content is used to train our models" for Free, Go, Plus and Pro, with "Opt-out available"; Business lists "No training on your business data by default." On any plan, keep customer names and contact details out. Clinics: Is ChatGPT HIPAA compliant?

Before you publish: a 6-point check

  1. Every price, promise and number matches your fact sheet.
  2. Every review is real, quoted as written, with its source.
  3. The headline repeats what the ad or post promised.
  4. The button says what happens next.
  5. On your phone, you tapped call, sent the form and booked a test slot.
  6. You wrote down the date and what changed, to compare leads before and after.

Example

Illustrative cases. Names and numbers are invented.

A roofer whose ads get clicks but few calls

His "Roof leak repair this week" ad, about 500 clicks a month at $6 USD each, opened his home page. ChatGPT summed that page up as "a roofing company that values craftsmanship": no leak repair, no "this week", no call button. He built a leak repair page with the ad's promise as its headline, cut the "lifetime warranty" the draft invented (his is 10 years), and compares calls for 4 weeks before and after.

A Calgary moving company's FAQ

The NEEDS ANSWER prompt flagged 3 questions the page never answered: piano moves, weekend rates and deposit refunds. The owner wrote the answers herself, including a weekend rate of $20 CAD more per hour, and added three Google reviews quoted word for word. She'll judge the change by quote requests over two months.

FAQ

Can ChatGPT look at my landing page from a link?

Don't rely on a link. Send a phone screenshot and the page text, so it reviews what visitors see first, including anything covering the button.

Can ChatGPT build the whole landing page?

Yes: ChatGPT Sites builds landing pages, in public beta on eligible Plus, Pro, Business, Enterprise and Edu plans. Confirm you can track calls and forms.

Will Google penalize a landing page ChatGPT wrote?

Not for using AI. Google says it can help research and structure original content, but mass-producing pages that add nothing can break its spam policy.

Different for you?

Keep learning

NextA/B testing for affiliate sites: what to test for more revenueHow to A/B test an affiliate site: setup, visitors needed, the page parts worth testing, and the disclosure and Amazon rules no test may break.

Ask us one question

One line is enough. We reply by email.

Sources

  1. OpenAI API docs: Prompt engineering
  2. OpenAI API docs: Images and vision
  3. ChatGPT pricing: plans, features and how content is used
  4. ChatGPT Sites: build and publish sites with ChatGPT
  5. Google Ads Help: Optimize your ads and landing pages
  6. Google Search Central: Guidance on using generative AI content on your website
  7. Google for Developers: About PageSpeed Insights
  8. Microsoft Learn: Clarity FAQ
  9. eCFR: 16 CFR Part 465, Rule on the Use of Consumer Reviews and Testimonials
  10. FTC: Final rule banning fake reviews and testimonials (2024)
  11. Justice Laws: Competition Act, section 74.02