A repeatable process for creating landing pages that moves from AI-generated concept through account manager alignment, designer polish, and developer build. This workflow was established during a Mark/Ben check-in focused on using ChatGPT for ad campaign analysis and landing page creation.
Use ChatGPT (or Claude) to produce an initial landing page concept. There are two starting points:
Starting from a live campaign:
- Download campaign data from Google Ads (campaigns, search terms, keywords) as Excel files
- Upload to ChatGPT and ask for analysis: what's working, what isn't, what the landing page should do
- Ask ChatGPT to generate a landing page spec and then an HTML/CSS mockup based on the campaign data
- This path produces higher-quality output because the AI can optimize the page for observed user behavior
Starting from a brief (no campaign yet):
- Provide ChatGPT with the brief and ask: "Is this enough information to make a good quality landing page?"
- If not, it will tell you what's missing — take those questions back to the account manager
- Once you have enough, ask for a mockup
In either case, ChatGPT will generate:
- A written spec (hero section, CTAs, benefits, social proof, lead form, FAQ, etc.)
- HTML/CSS code for a rough visual mockup you can preview in-browser
The mockup won't be beautiful, but it will be functional enough to communicate structure and content strategy.
Share the rough AI mockup with the account manager before involving design or dev. Be explicit that it's rough:
"Here's a super rough mockup — wrong colors, placeholder content — but does the structure and messaging make sense?"
The account manager will typically approve the concept or suggest adjustments. This step ensures alignment on content strategy before spending design time. The account manager may also take the spec to the client for input on things like calculator metrics, PDF assets, or CTA language.
Once the account manager approves the concept, send the designer:
- The AI-generated mockup (preview link or HTML file)
- Brand guide
- Approved images and assets
Ask the designer whether the HTML is useful to them directly (some can import it into Figma as a starting point). The designer produces a polished mockup with correct colors, fonts, and imagery.
Return the designer's mockup to the account manager for sign-off before build. This is the last checkpoint before developer time is committed.
Developer receives the approved designer mockup and builds the page. Best practice: build landing pages on the main site where possible, though this requires theme-level support for navigation-free layouts.
ChatGPT's output for the [1] commercial juicer campaign illustrates what the AI will typically recommend:
The AI will also specify color palette, typography, and mobile behavior.
| Task | Recommended Tool |
|---|---|
| Analyzing Google Ads data (Excel upload) | ChatGPT |
| Generating HTML/CSS mockups | ChatGPT |
| Browsing live URLs for page analysis | Claude |
| Creating hyperlinks from pasted URLs | Claude |
| General landing page critique | Either |
ChatGPT excels at code generation and structured data analysis. Claude handles web browsing and link creation more reliably. Both support projects/folders for organizing work by client. See also: [2].
Organize landing page work inside a ChatGPT Project (a persistent folder with memory across sessions). Create one project per client account so the AI retains context about the campaign, brand, and prior decisions.
If campaign analysis reveals something seriously wrong — budget being wasted, zero conversions, traffic going to the homepage instead of a landing page — flag it to the account manager immediately. Come prepared with the AI analysis and, ideally, a rough mockup of the fix. This positions the finding as a solution, not just a problem.
"I found an issue. Here's what ChatGPT says is wrong, and here's a rough mockup of what a better landing page could look like."