I Analyzed 508,034 Numerology Ratings to Improve an AI-Powered Product
I used AI to make numerology readings less generic. Then I analyzed 508,034 reader ratings.
I’ve been building Your Path Reading, a small numerology and astrology product.
The obvious way to use AI would be to enter someone’s birthday, generate a long mystical-sounding reading, and call it personalized.
That is also how you end up with generic content that could apply to almost anyone.
Instead, I used AI more like a research and editing tool:
• Compare hundreds of recurring reader reactions
• Identify which personality descriptions consistently resonate
• Find where readings feel vague, overly flattering, or inaccurate
• Separate meaningful patterns from generic “Barnum effect” language
• Turn numerology, Western astrology, and Chinese astrology into one coherent reading rather than three disconnected summaries
As part of that process, I analyzed 508,034 public ratings attached to numerology Life Path readings.
A few findings surprised me:
• 76% of readers rated their reading accurate
• Life Path 22 scored highest at 84.8%
• Life Path 5 scored lowest at 68.7%
• Master numbers averaged 83%, versus 75.1% for single digits
• Life Path 7 received the most ratings, which oddly fits its “outsider looking for an explanation” profile
This does not prove numerology is scientifically real.
But it does show that people respond very differently to different profiles. Some descriptions consistently make people feel understood. Others clearly miss.
That became the real product lesson for me:
AI is not most useful when it creates more content. It is useful when it helps you study feedback at a scale you could never process manually, then improve the product based on what people actually respond to.
The full analysis is here:
https://yourpathreading.com/numerology/is-numerology-accurate/
I’m still figuring out the commercial side of the product, but the research and product-development process has been one of the more interesting uses of AI I’ve worked on.
