One of my clients sells building materials — doors, tiles, flooring. Their
problem: customers didn't know what they needed. Product pages got traffic but
conversion was low.
I built a 5-question needs assessment that feeds into an AI recommendation
engine. Here's how.
The Architecture
User answers 5 questions
→ Form submits to Make.com webhook
→ Make.com calls ChatGPT API with answers + product catalog
→ AI generates 3 personalized recommendations
→ Automated email sent to user with product picks
→ Lead captured in CRM
All running on WordPress with Elementor for the frontend and Make.com for the
backend logic.
The Implementation
1. The Quiz (Frontend)
Built with a custom HTML/JS widget in Elementor. No plugin bloat — just a clean
multi-step form that posts to a Make.com webhook.
2. The AI Engine (Backend)
Make.com receives the form data, formats a prompt with the product catalog
context, and sends it to ChatGPT. The response comes back as structured JSON with
3 product recommendations + reasoning.
3. The Email (Delivery)
Make.com generates a conversion-optimized email with the 3 recommendations and
sends it via SMTP. The lead is simultaneously added to Pipedrive CRM.
Results
- Lead capture rate: 34% of quiz starters complete it
- Email open rate: 67% (because they're waiting for their results)
- Customer LTV increased 36% through better product matching
Key Takeaways
- WordPress is not "just a blog platform." With the right architecture, it handles AI integrations just fine.
- The quiz-to-email flow converts better than chatbots in this context. Users prefer answering structured questions over open chat.
- Make.com as middleware means zero changes needed on the WordPress side when we update the AI logic.
Full project details: L+B Bauhandel — E-Commerce + KI
Integration
More on how I approach web design for SMEs: Webdesign
Fürth
I build digital infrastructure for SMEs — websites that generate leads + AI
automation that processes them. Based in Fürth, Germany.
npc-agency.com
Top comments (0)