👨‍🚀 AI marketing hack: Prompt builds a website that converts

Hello AInauts,

Welcome to the new issue of your favorite newsletter! Today with practice, practice and more practice - promoting your own landing page, vibe-coding an app and having deceptively real voices tell a joke in dialect.

This is what we have in store for you today:

• 🚀 AI marketing hack: This prompt builds converting landing pages

• 🔥 TuneSpark - Our brute force vibe coding project...

• 🔊 ElevenLabs V3 - The most unfunny joke ... is awesome!

Ready? Let's go!

🚀 AI marketing hack: this new prompt builds killer converting landing pages

Imagine you have the perfect product or service - but no one is buying. Frustrating? Yes, really frustrating! ... Everyone who has a website wants to achieve more conversions with it.

The problem is often not the offer itself, but the fact that it is not positioned correctly. And this is where a landing page that sells for you 24/7 comes into play. It has one job: hold attention, build trust, trigger an action.

Here are a few examples of what the chatbots did with the prompt below from our German AI + Automation Accelerator page... we sent ChatGPT, Claude, Genspark and Manus into the race - ChatGPT disappointed, Genspark once again showed off its capabilities. Just click on the pictures and take a look at the pages yourself.

Simple prompts like "Create a landing page" do not work. Instead, you need conversion-optimized, context-rich prompts that are based on proven frameworks and best practices. With our new prompt, you can overcome the biggest hurdle in marketing and build a website that actually converts.

Just enter your homepage or product page and check out the results. Try it out - we look forward to your feedback!

## ROLE

You are an **elite conversion copywriter & funnel architect** with 10+ years of experience. You combine psychological sales triggers, modern UX principles, and data-driven conversion optimization with a wealth of experience from the best copywriters and their successful campaigns.

## ACTION

### Phase 1: Discovery & Analysis
1. **Ask the initial questions** and wait for complete answers.
2. **Analyze systematically**: offer, target audience, pain points, desired outcome, unique value proposition.
3. **Define conversion strategy** based on target audience psychology.
4. **Create images or image prompts** and insert them.

### Phase 2: Landing Page Engineering
Create a **high-converting landing page** using the following conversion frameworks:

**Psychological frameworks:**
- **AIDA 2.0**: Attention → Interest → Desire → Action (with micro-conversions)
- **PAS+**: Problem → Agitation → Solution → Proof → Action
- **Before & After Bridge**: Transformation journey with emotional reinforcement
- **Value Stack Formula**: Benefit building with anchor prices

**Modern conversion triggers:**
- Social proof (testimonials, reviews, logos)
- Authority positioning (expertise, certificates, media)
- Scarcity & urgency (limited time/quantity)
- Risk reversal (guarantees, right of return)
- Reciprocity (free, added value up front)
- Commitment & consistency (small yes steps)
- Loss aversion (what do you lose without the offer?)

**Design & layout:**
- Responsive grid – column grid, max. line length 70 characters
- Image placeholders – standard ratio 16:9, min. 1200 px width
- If possible: link to image URLs/videos on the source page
- Contrast – WCAG AA for text/BG, buttons ≥ 4.5:1
- White space – min. 24 px vertical between sections, 16 px internal padding
- Typography – sans-serif system font stack, H1 2.5 rem, body 1 rem, line height 1.6
- Colors – primary action color max. 2 × per screen, neutral background tones (#F9FAFB, #FFFFFF)
- Icon style – uniform (e.g., Lucide), size 24 × 24 px
- Error tolerance – display SVG placeholders if images are not available (onerror → fallback SVG)
- Above the fold – no scroll jumps, CLS < 0.1
- Interaction feedback – hover states + active states for all CTAs
- Accessibility checks – form labels, ARIA landmarks, ...

### Phase 3: Creation, review, and optimization
- Review the result – and make it 10 times better! 
- Then assign a rating on a scale of 1-10 and continue with phase 3 until you achieve a rating of at least 9/10.
- Output the formatted landing page in ChatGPT Canvas, Claude Artifacts, or similar.

## FORMAT / STRUCTURE

Return the landing page as a finished page – without any placeholders for sections, format, images, links, or CTAs. Instead, use clear, concrete wording in brackets, e.g., “(Hero image: young marketer in front of dashboard)”.

### LANDING PAGE BLUEPRINT
1 paragraph = 1-2 sentences, total ≥ 300 words. Do NOT output section titles, but generate corresponding content.

| Section | Goal | Conversion elements | Length |
|---------|------|-------------------|-------|
| **🎯 Hero fold** | Immediate attention + clarity | Headline (6-12 words), subheadline, primary CTA, hero visual, trust logos | 1 screen |
| **😤 Problem intensification** | Intensify pain + urgency | Dramatize the pre-state, “What happens if nothing changes?” | 2-3 paragraphs |
| **✨ Transformation Promise** | Hope + vision | Visualize the post-state, emotional benefits | 2-3 paragraphs |
| **🎁 Value Proposition** | Clearly communicate USP | “Only we can...”, unique selling points | 1 paragraph |
| **💎 Benefits Arsenal** | Translate features into benefits | 5-8 benefits with micro-outcomes, icons/emojis | List/grid |
| **📦 Offer Breakdown** | Transparency + value creation | Detailed list of all components, individual values | Table/list |
| **🏆 Social Proof Stack** | Trust + credibility | 3-5 testimonials, logos, figures, awards | 2-3 blocks |
| **🔬 Proof of Concept** | “Why does it work?” | Simple explanation of the methodology/system | 1-2 paragraphs |
| **🛡️ Risk Reversal** | Reduce purchase resistance | Guarantees, return policy, “What's the worst that could happen?” | 1 paragraph |
| **❓ FAQ & Objection Handling** | Remove any remaining doubts | 5-7 frequently asked questions + objections, including price | Q&A format |
| **💰 Pricing & Urgency** | Push the purchase decision | Price anchor, comparison, limitation, bonus for immediate purchase | Pricing box |
| **🚀 Final Push CTA** | Last conversion attempt | Summary, strongest CTA, P.S. with urgency | Call-to-action block |

### INITIAL QUESTIONS (Quick-Start Briefing)
**What is your offer?** Describe it or enter the URL.  
*(Optional: informal or formal address, target group, problem, style, images, link call-to-action/checkout, colors, ...)*
*Do not ask any further follow-up questions! Make assumptions where necessary.*

### Conversion optimization
- **Mobile-first**: Smartphone-optimized (>60% traffic)
- **Readability**: Flesch score >60, short sentences
- **CTA density**: One CTA every 1-2 screen lengths
- **Visual hierarchy**: Follow the F/Z pattern
- **Clarity**: No technical terms without explanation
- **Emotion**: 70% emotional, 30% rational
- **Specificity**: Concrete figures instead of vague statements
- **Urgency**: Time pressure without manipulation
- **Authenticity**: Honest claims, realistic promises

### RESULT
A publishable, scientifically sound, psychologically optimized landing page with graphics/images that has been proven to achieve higher conversion rates than standard templates thanks to proven frameworks and modern UX principles.

🔥 TuneSpark - Our Brute Force Vibe Coding Project...

A few weeks ago, we wrote a post about the latest, greatest Suno 4.5 update - and then had a "little" idea that we wanted to implement.

We have tunes in our ears all day and of course have already compiled an extensive list of favorite songs on Spotify. This gave rise to the idea of building an app using vibe coding.

The goal: to put your favorite Spotify songs in and get new song prompts out - to generate fresh tracks that exactly match your vibes: TuneSpark!

Very simple: From the idea to the first app launch …

We like to test a lot and have already talked about Vibe Coding several times. And that's why we thought it was a great idea - but sometimes it's easier said than done.

• The first step is always very simple: Google Gemini quickly created a detailed specification for the app based on a few sentences.

• We then uploaded this PRD ("Product Requirements Document") one-to-one in parallel to Replit, Lovable, Firebase and Bolt and simply let the Vibe coding apps run.

• Unfortunately, Bolt shot itself in the foot here. After 15 unsuccessful attempts to solve a simple error, we wanted to spare our nerves and put it aside for the time being.

• With Lovable, the focus was more on design than function in the first pitch.

• Replit has built a cool, almost functional app with demo data.

"Almost there!" ... at least that's what we thought.

After the initial euphoria, however, the stumbling blocks appeared.

• Replit simply didn't want to do the Spotify API integration in such a way that our own songs were actually available for selection.

• Unfortunately, we quickly reached the limit of available prompts with Lovable (5 per day... we've long since exhausted the 100 per month) and had to put it on hold for the time being.

• That left us with the relatively new Google Firebase. Firebase impressed us very positively right from the start and built a cool, functional app!

• The problems started when we wanted to publish it. We suddenly had to assign some strange authorizations in Google Cloud, which still didn't work as they should have…

First attempt: failed! Time for a reset...

At the first attempt, we didn't get straight to the goal with any app - although it felt like 80% of the requirements were already in place.

What do we do now? We fed the original specification and the entire conversation with the bugs to Gemini 2.5 Pro again - with the request to write a much more detailed specification with a precise process for the Spotify API connection.

No sooner said than done! We then packed this new specification into a new project at Replit and Google Firebase and used it to generate version 2.

Brute Force: A second attempt at our own app!

And with the knowledge gained from the failed attempts, we were a big step ahead right from the start. The Spotify API integration was suddenly no longer an insurmountable hurdle, but was completed after a short back and forth, and the cloud integration now also ran smoothly.

A simple prompt always served us well during testing:

Do you need to fix anything else before I start testing it? Take a deep, hard look at the code and make sure everything works properly. I'm counting on you!

Of course, we wanted to include a few more features - and then we actually managed to release the app for testing. Voilà!

Try it out: Either with demo data or directly with your own favorite Spotify songs as inspiration for new AI tracks with the music generator of your choice.

P.S. With the very latest models such as Claude 4 and Gemini 2.5 Pro, everything has gotten even better in recent days - the apps make fewer mistakes or iron them out independently.

🔊 ElevenLabs V3 - The least funny joke ... is awesome

Ok, this issue has gotten a bit long... but the latest version 3 of ElevenLabs is just so awesome that we want to show you an example.

Just listen to it - and then clone your own voice on ElevenLabs or play around with a canned voice.

It's official: AI-generated voices can no longer be distinguished from real ones!

We made it! But no need to be sad. The AInauts will be back soon, with new stuff for you.

Reto & Fabian from the AInauts

P.S.: Follow us on social media - it motivates us to keep going 😁!
X, LinkedIn, Facebook, Insta, YouTube, TikTok

Your feedback is essential for us. We read EVERY comment and feedback, just respond to this email. Tell us what was (not) good and what is interesting for YOU.