One Founder Video Lifted Conversion Rate 33% — Here's the Claude Code Landing Page Stack Behind a $1.2M Business
A founder video moved CVR from 10% to 15%. Video testimonials cut Google Ads CPA 7x. Here's the full Claude Code stack that powers it.
A Single Founder Video Moved CVR From 10% to 15% — Here’s What That Actually Means
Adding one founder video to a landing page lifted conversion rate from 10% to 15%. That’s a 33% increase in leads with zero other changes. The same builder also added video testimonials to a separate Google Ads campaign and watched cost-per-conversion drop from $200 to $30 — a 7x improvement. These aren’t projections from a case study PDF. They’re from Jono Catliff’s own analytics, from a wedding DJ and photography business he built and sold after spending $171,000 on Google Ads.
The numbers are specific enough to be worth taking seriously. And the tooling he used to build, test, and deploy the pages is entirely free.
Why Video Converts When Nothing Else Moved the Needle
The instinct when a landing page underperforms is to fix the design. Better hero image, tighter headline, different color on the CTA button. Catliff’s framing is blunter: beautiful websites make you zero dollars. Design is a binary trust signal — either the page looks credible enough to continue, or it doesn’t. Past that threshold, aesthetics stop mattering.
What does matter is whether the visitor trusts the person they’re about to hire. A faceless brand with polished copy is still a faceless brand. A 45-second video of the founder — not pitching services, just being a recognizable human — closes that gap faster than any headline variant.
Plans first. Then code.
Remy writes the spec, manages the build, and ships the app.
The video testimonials result is even more interesting. Catliff’s framing: people might not even watch them, but the fact that they’re there provides social proof. The presence of video testimonials signals that real customers exist and were willing to go on camera. Written testimonials don’t carry the same weight anymore — they’re too easy to fabricate. (He mentions thispersondoesnotexist.com in passing, which is a fair point about stock-photo testimonials.) Video is harder to fake, and visitors seem to price that in even when they skip past it.
The practical implication: if you’re running Google Ads and your cost-per-conversion is stuck, the problem might not be your targeting or your bid strategy. It might be that your landing page has no video.
The Stack That Makes This Buildable Without a Dev Team
The tooling here is worth cataloguing precisely because it’s all free and requires no coding background.
The build environment: Antigravity IDE is a Google product — a free desktop app that functions as a coding workspace. Claude Code runs inside it as a plugin. You install it from the left sidebar, search for “Claude Code,” and you’re building. The claude.md file that Catliff uses to configure Claude’s behavior for web projects is downloadable from his free School community. Think of it as an onboarding document for your AI collaborator — you’re telling it how to behave before it writes a single line.
Design input: Rather than starting from a blank prompt, he pulls design references from Dribbble. Search “landscaping website,” screenshot something you like, upload it to Claude with a prompt like “build me a one-page website for my landscaping business and replicate this design.” Claude handles the rest. The output isn’t pixel-perfect, but it’s close enough to iterate from.
Deployment: GitHub repo → Vercel import. Select the Next.js preset, add your environment variables (the PostHog API key that gets generated during setup), and the site is live in about 60 seconds. Both GitHub and Vercel are free at this scale.
Analytics and testing: PostHog handles A/B testing, session recordings, and heatmaps. Installation is via a PostHog wizard that runs in the terminal — you paste one line, hit enter twice, and it connects your project to PostHog automatically. The one non-obvious setup step: you need to add authorized URLs in PostHog’s settings under Web Analytics before experiments will track correctly. The conversion event to track is a page load on the thank-you page, not a form submit event — it’s simpler to implement and just as accurate.
If you’re building the kind of multi-model workflow that sits behind a landing page — lead scoring, automated follow-up sequences, CRM routing — platforms like MindStudio handle that orchestration layer: 200+ models, 1,000+ integrations, and a visual builder for chaining agents without writing the glue code yourself.
The Conversion Architecture Behind the 20% Rate
The video is the headline finding, but it sits inside a broader page architecture. Understanding the full structure matters because the video alone won’t get you to 20% — it’s one component in a system.
Copy: Less is more, with one exception. Catliff’s rule is that visitors read about 10% of your page text, so you want to control which 10% they see. Short headlines, benefit-focused subheads, minimal body copy. The exception is SEO pages, which need 1,500+ words — his workaround is accordion/dropdown sections that hide the text unless clicked.
- ✕a coding agent
- ✕no-code
- ✕vibe coding
- ✕a faster Cursor
The one that tells the coding agents what to build.
No navbar, no footer (for ad traffic): Every link that isn’t “call us” or “fill out the form” is a distraction. If you’re paying $10 per click, you don’t want that click going to your About page. Kill the nav. Kill the footer. One page, two actions.
Call-to-action placement by device: On desktop, CTAs are sprinkled throughout the page — wherever a visitor might decide they’ve read enough. On mobile (64% of your traffic, per the cited stat), there’s one CTA pinned to the bottom of the screen at all times. The mobile visitor never has to scroll back up to convert.
The zipper approach for scale: One homepage plus a matrix of location/service landing pages. City × service = a page. “Tree trimming Toronto,” “tree trimming Manhattan,” “tree trimming the Bronx.” Each page matches the exact search query that brought the visitor there. The /landing-page-generator [city] [service] [traffic-source] skill in Claude Code generates these on demand — you define it once, then call it with different parameters. For SEO, the guidance is to ramp slowly; spiking 1,000 pages in a day will trigger a Google penalty.
Speed: The data Catliff cites is stark. At 1 second load time, average CVR is 3.05%. At 2 seconds, it’s roughly half that. At 4 seconds, you’re at about 20% of the 1-second baseline. The fix is Google Lighthouse, which is built into Chrome DevTools. Run an analysis, copy the full report, paste it into Claude with “optimize for page load speed and all device types.” Catliff’s demo showed a jump from 90% to 95% performance score in one pass. Keep iterating until you hit 100%.
For teams building more complex full-stack applications where the spec itself needs to be the source of truth, Remy takes a different approach: you write an annotated markdown spec and it compiles a complete TypeScript backend, SQLite database, auth, and deployment from it. The generated code is derived output — the spec is what you maintain. That’s a different abstraction level than Claude Code’s interactive build loop, but relevant when you’re moving from landing page to actual application.
What the A/B Testing Actually Tells You
The split test setup in PostHog is straightforward: create an experiment, split traffic 50/50 (or more variants), set the primary metric to the quote_submitted page load event, launch. The subtlety is in the interpretation.
Catliff’s point is that nobody — not you, not Claude, not Google — can tell you definitively what will get you to 20% conversion rate. The only way to know is to test. He ran 40 to 50 landing page variants before finding the formula that held for years. The data doesn’t lie; your intuitions about what visitors want usually do.
The session recordings in PostHog add a qualitative layer. You can watch real visitors navigate your page and see where they get confused, where they drop off, what they click that you didn’t expect them to click. It’s less statistically rigorous than the split test, but it surfaces anomalies that aggregate data hides.
Not a coding agent. A product manager.
Remy doesn't type the next file. Remy runs the project — manages the agents, coordinates the layers, ships the app.
One practical note on PostHog’s cookie behavior: always test your A/B routing in an incognito tab. PostHog sets a cookie that locks a visitor to one variant. In your regular browser, you’ll always see the same page. Incognito clears the cookie, so each new tab simulates a fresh visitor being randomly assigned.
Speed-to-Lead: The Conversion Multiplier After the Form Submit
The landing page is only half the system. What happens after someone fills out the form matters as much as what got them to fill it out.
The data point Catliff cites: calling a lead within one minute of form submission produces 4x more sales than calling later. The mechanism is psychological — someone who just submitted a form is in “get this done” mode. They haven’t started comparing competitors yet. They haven’t talked themselves out of it. Call them in 60 seconds and you’re the only option they’ve seriously considered.
The automation: form submit → webhook → CRM → automated outbound call in approximately 10 seconds. Twilio is the VoIP backbone for most CRMs. The demo in the video shows a call coming through with “Hey Jono, you have a new lead — press any key to connect.” Press a key, and you’re live with the prospect. The whole sequence from form submit to live call is under 15 seconds.
This is worth building even if you’re not running Google Ads. If you’re doing SEO and generating organic leads, the same speed advantage applies. The lead doesn’t know how they found you; they just know whether you called them before they moved on.
For anyone building the webhook-to-CRM piece, the tooling options are Zapier, Make.com, n8n, or direct CRM API integration. The specific tool doesn’t matter much — what matters is that the call happens automatically, not when you happen to check your email.
The Honest Accounting
The 20% conversion rate is real, but it took 7 years and 40-50 landing page tests to get there. The founder video is the single highest-leverage change Catliff identifies — 10% to 15% CVR, 33% more leads, from one addition. The video testimonials are the second — $200 to $30 CPA, 7x improvement, from nine videos filmed on phones.
Neither of these requires Claude Code. You could add a founder video to any landing page today. The Claude Code stack — Antigravity IDE, claude.md configuration, Dribbble for design references, PostHog for testing, Vercel for deployment — makes it faster to build and iterate, but the underlying insight is about human trust signals, not tooling.
The tooling matters for scale. If you’re building a city × service matrix of landing pages, the /landing-page-generator skill approach is the difference between doing that manually for weeks and doing it in an afternoon. If you’re running split tests across 10 variants, PostHog’s free tier handles it without a data team.
Building animated 3D landing pages with Claude Code is one way to push the design ceiling higher once the conversion fundamentals are in place. But the conversion fundamentals come first. A 45-second founder video, filmed at a WeWork on a Saturday, moved the needle more than any design work Catliff describes.
Remy doesn't build the plumbing. It inherits it.
Other agents wire up auth, databases, models, and integrations from scratch every time you ask them to build something.
Remy ships with all of it from MindStudio — so every cycle goes into the app you actually want.
The uncomfortable version of this finding: most landing pages are missing the thing that matters most, and it’s not a technical problem. It’s a willingness to be on camera.
For teams already comfortable with Claude Code’s build loop, the effort level settings are worth understanding when you’re iterating on Lighthouse optimization passes — higher effort on the analysis step, lower effort on the mechanical fixes, saves tokens without sacrificing output quality. And if you’re building the content layer around your landing pages, the skill-based content machine approach applies the same /skill [parameters] pattern to content production that the landing page generator applies to page creation.
The stack is free. The insight is old. The combination is what’s new.
