Happy Thursday!
No guest this week. Just me, rambling through the exact landing page system I've been running for the last few weeks, because it came up on two separate calls the other week and I figured you'd want the whole thing. One call was a $300M supplement brand trying to crack Meta again. The other was a $500K brand paying an agency to build them better pages.
Same problem on both ends, and the fix is the same too: if you're driving paid traffic to a page you haven't perfected yourself, that's on you.
Stop Building Landing Pages by Hand
The basic lifecycle of a landing page: idea → strategy → research → copywriting → wireframe → UI design → development → QA → iteration.
Keep everything up to the wireframe in-house. Why? Nobody knows your brand, your customer, or your objections better than you, and the tools are now good enough that you barely have to lift a finger to do it yourself.
Keep everything up to the wireframe in-house.
Proof it works: the $500K brand's founder showed me her agency's Figma next to a page she built herself in Claude Code. Different angle, different copy, different variant-selection flow, different everything. Hers was better on every count. And last week I stayed at the office until 4am and shipped nine advertorials and two influencer landing pages in about five hours, learning the process as I went.
Listen here: YouTube | Spotify | Apple
What I pulled from this one:
- Strategy, research, copy, and wireframe stay in-house. An agency is great at executing a page. It has never tried your product, never read your CS inbox, and never watched your ad comments. If you're spending $10K or $10M a month on Meta and outsourcing the thinking, the bad page is your fault.
- Treat research like a data pipeline. Reviews, ad comments, CS tickets, DMs, creator feedback, TikTok, Reddit, YouTube, forums, e-retail reviews. Pipe every source into your agent, tell it to keep watching, and write every pattern into one doc. Give it more context than feels reasonable. That's how it starts thinking like you.
- Talk the page into existence. Paper sketch or a voice memo becomes a brief in Codex or Claude Code, which becomes a wireframe in Manus. No Figma rectangles, no back-and-forth with a designer on layout. The wireframe shows up already snapped into place.
- Pay a human for UI and dev. AI wireframes are excellent. AI UI design is still a month or two out. A premium-looking page is a trust signal on par with a great lifestyle photo on your PDP, so unless you already have a modular design system built, put real people on this step.
- QA and iteration go back to the agent. Stress-test across iPhone sizes, Android, Wi-Fi and cell, every laptop width. Check pixels fire, advertorial domains are whitelisted, popups and floating CTAs behave. Then pipe Clarity heatmaps and GA4 back into your agent so the second wireframe is smarter than the first.
At a glance
Who owns each part of the page?
Keep ownership clear before the first wireframe.
| Part of the work | Responsibility described in the episode |
|---|---|
| Strategy and research | The founder and team provide the customer context, objections, and source material. |
| Copy and wireframe | Keep the thinking in-house. Use a sketch or voice memo to brief the tools. |
| UI and development | Bring human specialists into the design and implementation. |
| QA and iteration | Use the agent to help test the page, then feed observed behavior into the next revision. |
That's all for today
Build one of these pages and DM it to me on Twitter. I've been giving feedback over DMs and having a lot of fun with it. And if you're not building one, tell me why. I want to know what's actually stopping you, because this is going to get simpler, and the brands that learn it now won't be the ones scrambling later.
It's Thursday, and it's all downhill from here. Get some rest, drink some water, and I'll see you Sunday.
Nik