Creator Intelligence Platform Development

I am setting up a web-based “Creator Intelligence” platform that lets Amazon KDP authors, Amazon FBA sellers, YouTube creators and broader ecommerce merchants run head-to-head A/B tests on the assets that matter most to them. Users will upload two variants—cover A vs cover B, thumbnail A vs thumbnail B, product photo A vs product photo B, and so on. Once enough test traffic or panel responses are collected, the system must declare a clear winner, display supporting data, and generate an AI-written insight report that explains why one option outperformed the other.

Scope of what must be testable right out of the gate
• Book covers, titles, subtitles, descriptions and A+ content for KDP authors
• Product images, packaging concepts, listing titles and brand names for FBA sellers
• YouTube thumbnails, video titles and channel branding for creators
• Product photos, store banners, product names and ad creatives for ecommerce stores

Core functionality I need built
• Simple drag-and-drop interface to upload the two variants of any asset type
• Choice of traffic source: internal panel, external survey link, or live visitor split (toggle per test)
• Statistical engine that tracks impressions, clicks and conversions, then calculates significance automatically
• AI layer (OpenAI or similar) that reads the quantitative results, highlights the key differences, and writes a short narrative summary for the user
• Dashboard showing test progress, confidence level, and final “winning” asset with exportable charts
• User segmentation so an account can run multiple tests at once across the four creator groups above
• Clean, responsive front-end (React or similar) and a scalable back-end (Node/Express, Python or comparable stack) with secure authentication and payment integration for subscription tiers

Acceptance criteria
1. A first test can be created, launched and concluded end-to-end without manual database edits.
2. The AI summary references the actual data points (e.g., “Cover A achieved 62 % higher click-through rate with 95 % confidence”).
3. Results dashboard exports to PDF/CSV in one click.
4. Page load time stays under 2 s for typical image assets (<5 MB each).

If you have previous experience with split-testing tools, survey panel integrations, or AI content generation, that will speed things up—but solid full-stack skills and a clear plan for the statistical component are what matter most. Let me know which frameworks you would choose and how you would tackle the significance calculations and AI analysis layer, and we can move forward.